feat!: modularize stemmer models and release infrastructure

Move bundled stemmer dictionaries from the core artifact into independently
versioned model modules. Add model discovery and explicit model-loading APIs,
a standard model aggregate, a model BOM, and dedicated model and catalog
release workflows.

Add full PoliMorf integration, model provenance and licensing validation,
streaming model-input verification, strict dependency verification, consumer
resolution tests, Configuration Cache compatibility, and expanded JMH,
quality, documentation, and release checks.

Upgrade the CycloneDX and JMH Gradle plugins and remove Gradle 10 and Java
compiler deprecations.

BREAKING CHANGE: The core Radixor artifact no longer contains bundled stemmer
dictionaries. Applications must add the required model artifacts, the standard
model aggregate, or model dependencies managed through the Radixor model BOM.

docs/benchmarks/reference/linguistic-quality.md

@@ -4,7 +4,9 @@ This evaluation measures agreement between the relation predicted by a stemmer a
 
 ## Scope and fair-comparison rules
 
-The authoritative Radixor language universe is the reconciled set of `stemmer.gz` resources under `src/main/resources` and `StemmerPatchTrieLoader.Language`. Radixor is evaluated for every reconciled language. A third-party adapter is evaluated only for languages supported by its tested implementation and having a compatible Radixor dictionary; unsupported combinations are absent rather than assigned zero quality.
+The authoritative Radixor language universe is the reconciliation of registered default model descriptors and `StemmerPatchTrieLoader.Language`. Radixor is evaluated for every reconciled language. Optional models are separate comparison rows. A third-party adapter is evaluated only for languages supported by its tested implementation and having a compatible Radixor dictionary; unsupported combinations are absent rather than assigned zero quality.
+
+Model identity is part of the candidate identity. Default Polish means `pl-pl-unimorph`; optional PoliMorf means `pl-pl-polimorf`. Results for those inputs must not be combined or relabeled, and historical snapshots cannot acquire a newer model identity retroactively.
 
 Within one language and dictionary mode, every adapter receives the same original included forms. Exact duplicates are removed only within one dictionary row. Identical surface forms in different rows remain distinct entries. Candidate strings use exact `String.equals`, with no evaluation-only lowercasing, normalization, accent removal, or gold-label-aware selection. Adapter preprocessing and lifecycle match the JMH comparison path.
 

docs/benchmarks/reference/methodology.md

@@ -1,6 +1,6 @@
 # Benchmark Methodology
 
-The stemmer comparison suite measures Radixor and Java stemmers on the same language and deterministic Radixor dictionary-derived data. Published Radixor rows in this refresh use contracted compiled patch tries, where uniform preferred-command subtrees are collapsed into accepting leaves before the trie is frozen for lookup. For each language, the bundled dictionary resource stores the expected root as the first tab-separated field on a line and its surface forms on the same line. Every single-token field on that line can therefore be paired with the same expected root.
+The stemmer comparison suite measures Radixor and Java stemmers on the same language and deterministic Radixor model dictionary-derived data. Published Radixor rows in this refresh use contracted compiled patch tries, where uniform preferred-command subtrees are collapsed into accepting leaves before the trie is frozen for lookup. For each language, the registered default model resource stores the expected root as the first tab-separated field on a line and its surface forms on the same line. Every single-token field on that line can therefore be paired with the same expected root.
 
 Published stemmer comparison results must come only from benchmark classes matching `.*StemmerComparisonBenchmark.*`. Internal `FrequencyTrie*` microbenchmarks are not part of those results.
 
@@ -33,7 +33,7 @@ Radixor is measured over dictionary tokens from its own resources: lower-case wi
 
 Lucene TokenFilter paths include required normalization in the measured pipeline. Examples include lower-case normalization for filters requiring lower-case input, German normalization before German light/minimal stemming, and Persian decimal, Arabic, and Persian normalization before Persian stemming. No ASCII folding is applied to Czech or Polish paths, because those Lucene stemmers are diacritic-aware or dictionary/table-backed for those languages. TokenFilter throughput methods materialize each emitted `CharTermAttribute` as a `String` before passing it to the JMH `Blackhole`, so output consumption is easier to inspect and closer to the direct stemmer methods.
 
-For right-to-left Radixor languages, patch application uses the traversal direction stored in trie metadata. This is required because static backward patch application is not correct for all bundled languages.
+For right-to-left Radixor languages, patch application uses the traversal direction stored in trie metadata. This is required because static backward patch application is not correct for all registered language models.
 
 ## Quality Metric
 
@@ -57,3 +57,4 @@ Morfologik can emit multiple terms for one input token. The quality benchmark us
 Quality reports use JMH auxiliary counter rows. Exact-root accounting is deterministic for a fixed corpus and stemmer, so repeated measurement samples duplicate the same counters; documentation uses the counter ratios and does not interpret quality benchmark timing scores.
 
 Pairwise over-stemming, under-stemming, candidate-aware policies, balanced accuracy, and partition comparison are a separate analytical evaluation. See [Linguistic Quality Methodology](linguistic-quality.md); exact-root accuracy must not be interpreted as the complement of pairwise under-stemming.
+Default rows use `Language.defaultModelId()`. Optional variants require a separate model field; `pl-pl-unimorph` and `pl-pl-polimorf` must never share an ambiguous Polish label. The benchmark runtime receives each resource exactly once from its individual model JAR through direct JMH runtime dependencies. See [Model Selection and Loading](../../model-selection-and-loading.md).

docs/benchmarks/reference/reproducibility.md

@@ -20,7 +20,8 @@ The CSV contains raw TP, FP, FN, and TN counts; raw over/under numerators and de
 ./gradlew publishStemmingQualityDocumentation
 ./gradlew verifyStemmingQualityDocumentation
 ./gradlew test
-mkdocs build --strict
+./gradlew prepareMkDocsSource
+mkdocs build --strict --config-file build/mkdocs/mkdocs.yml
 ```
 
 `stemmingQuality` performs the expensive complete evaluation and is intentionally not attached to `test` or `check`. It prepares JMH third-party dependencies automatically and writes:
@@ -34,6 +35,26 @@ Audit mode is enabled with `-PstemmingQualityAudit=true`. Language, stemmer, dic
 
 `publishStemmingQualityDocumentation` validates the complete build CSV, copies a versioned documentation snapshot, and replaces only marked generated sections. `verifyStemmingQualityDocumentation` re-renders from the checked-in snapshot and fails on changed values, ordering, missing pages, duplicate keys, arithmetic inconsistencies, policy violations, or stale sections.
 
+The model catalog and rendered site are build outputs under `build/`. They are generated for publication and are never maintained in Git.
+
+For new measurements, record language, stable model ID, model artifact version, descriptor checksum, source dictionary identity/version, core revision, and benchmark configuration. JMH resolves the required default models and optional PoliMorf directly from their individual model JARs; these benchmark-only dependencies are not transitive to ordinary users.
+
+Current model descriptors also record the official repository, dataset, license, attribution,
+verification date, transformations, and source-revision status. Exact historical revisions were
+not recorded for the legacy UniMorph imports; that limitation is disclosed with
+`not-recorded-in-legacy-import` rather than reconstructed. Future imports must record the exact
+upstream revision and source-archive checksum. This reproducibility limitation does not replace or
+weaken the packaged license and attribution requirements.
+
+Each UniMorph-derived model artifact carries its own notice with the canonical CC BY-SA 3.0 URI,
+upstream attribution, transformations, ShareAlike statement, and Leo Galambos contribution notice.
+The full CC legal text is not duplicated or presented as a root-project license. PoliMorf retains
+its separately packaged BSD-2-Clause license.
+
+For a future full PoliMorf measurement, also record the startup heap separately from benchmark parameters. Complete runtime construction is currently verified with a dedicated 6 GiB maximum heap; this limit is neither a retained-trie measurement nor a setting applied to ordinary JMH runs.
+
+The Pages workflow publishes that staged documentation together with Javadoc, JUnit, PMD, JaCoCo, PIT, representative JMH, SBOM, optional dependency-check output, badge metadata, and retained build history. Its filesystem merge explicitly preserves the `builds/` tree in the separate `gh-pages` publication branch, so documentation regeneration cannot erase durable report URLs.
+
 ## Performance benchmark reproduction
 
 The JMH comparison command family is:
@@ -46,7 +67,7 @@ The exact JMH configuration, hardware, operating system, and JDK captured for th
 
 ## Recorded and unavailable provenance
 
-The performance documentation records its 2026-07-06 environment, JDK 25.0.3, operating system, and hardware. The quality CSV records the evaluated identifiers and counts but does not embed the Radixor Git revision, generation date, JDK, operating system, dictionary content hash, or immutable upstream revisions for every downloaded source. These fields are explicitly unavailable for this snapshot and are not reconstructed from filesystem timestamps.
+The performance documentation records its 2026-07-06 environment, JDK 25.0.3, operating system, and hardware. The quality CSV records the evaluated identifiers and counts but does not embed the Radixor Git revision, generation date, JDK, operating system, model ID, dictionary content hash, or immutable upstream revisions for every downloaded source. These fields are explicitly unavailable for this historical snapshot and are not reconstructed from filesystem timestamps. In particular, the snapshot predates the optional PoliMorf integration and must not be relabeled as `pl-pl-polimorf`.
 
 Dependency versions that are reproducible from repository configuration include Apache Lucene 10.5.0, Morfologik 2.1.9, the Ukrainian dictionary artifact 4.9.1, and JMH 1.37. Other upstream branches or downloaded dictionary revisions should be pinned and embedded in a future result schema.
 
@@ -59,3 +80,4 @@ Audit reports preserve original multilingual forms and identify high-contributin
 ## JMH badge compatibility
 
 The quality documentation generator does not invoke JMH, change JMH result formats, or modify badge tooling. Existing JMH result paths and historical badge-compatible inputs remain independent. The repository currently publishes coverage and mutation badge metadata and retains JMH TXT/CSV artifacts as documented in [Environment and reports](environment.md).
+See [Model Selection and Loading](../../model-selection-and-loading.md), [Stemmer Models](../../stemmer-models.md), and the generated [model catalog](../../stemmer-model-catalog.md) for current model identities.

docs/benchmarks/reference/tested-stemmers.md

@@ -4,7 +4,7 @@ The JMH adapter registry is authoritative for evaluated implementations and lang
 
 | Family or implementation | Upstream / attribution | Tested version or revision | Evaluated scope | Output capability and adapter behaviour | Interpretation notes |
 | --- | --- | --- | --- | --- | --- |
-| Radixor | Egothor / Radixor project | Current repository revision; exact revision was not embedded in the quality CSV | All 20 reconciled Radixor dictionary languages; 19 have benchmark pages | Deterministic preferred patch via `get`; ranked distinct alternatives via `getAll`; primary is always included | Dictionary-derived compiled patch trie. Quality depends on dictionary coverage and annotation. |
+| Radixor | Egothor / Radixor project | Current repository revision; exact revision was not embedded in the quality CSV | All 20 reconciled default model languages; 19 have benchmark pages | Deterministic preferred patch via `get`; ranked distinct alternatives via `getAll`; primary is always included | Model-dictionary-derived compiled patch trie. Default rows use each language's stable default model ID. |
 | Apache Lucene language stem filters | Apache Lucene project | 10.5.0 | Adapter-declared language-specific subsets | TokenFilter lifecycle and language normalization match JMH; normally single-output | Light, minimal, possessive, and language stem filters deliberately implement different scopes. Narrow scope is not a defect. |
 | Apache Lucene SnowballFilter | Apache Lucene project using Snowball algorithms | Lucene 10.5.0 | Snowball-supported subset of Radixor languages | Single primary token emitted through the Lucene TokenFilter path | Includes TokenStream overhead and required normalization. |
 | Official Snowball Java | Snowball project | Repository preparation downloads the configured upstream Java distribution; an immutable revision was not recorded in the quality CSV | Same-language adapter subset | Direct generated Java API; single output | Rule-based suffix algorithms provide broad baselines rather than dictionary-root guarantees. |
@@ -24,6 +24,7 @@ Candidate sets are non-null, non-empty, contain the deterministic primary output
 
 ## Coverage fairness
 
-Radixor coverage is derived independently from its resources and language enumeration. Third-party coverage is the intersection of that universe with actual adapter support. Absence therefore means “not supported or not configured for this language,” not “zero quality.” Consult each language page for the exact evaluated rows.
+Radixor coverage is derived from registered default descriptors reconciled with language enumeration. Third-party coverage is the intersection of that universe with actual adapter support. Absence therefore means “not supported or not configured for this language,” not “zero quality.” Optional `pl-pl-polimorf` is a separate model row and does not replace default `pl-pl-unimorph`. Consult each language page for the exact evaluated rows.
 
 Project authors and organizations are named only where repository configuration or source notices establish attribution. No broader authorship or license claim is inferred when metadata was not captured.
+The JMH runtime configuration directly includes optional models needed for controlled comparisons; ordinary users do not receive these benchmark-only dependencies transitively. Historical rows retain their original model inputs. See [Model Selection and Loading](../../model-selection-and-loading.md).