feat: add AI-based security suggestion engine and CLI integration

Introduce new package org.egothor.methodatlas.ai providing AI-assisted
classification of JUnit tests for security relevance.

Key changes:
- add AI suggestion engine, provider abstraction, and provider clients
  (OpenAI-compatible, Ollama, Anthropic)
- implement strict JSON prompt/response contract and taxonomy handling
- integrate AI enrichment into MethodAtlas CLI output (CSV and plain
  modes)
- add configuration via AiOptions and CLI flags
- add comprehensive JUnit + Mockito test coverage for AI components
  and CLI integration scenarios
- add realistic test fixtures for security-related test classes
- update Gradle configuration for Mockito agent support on JDK 21+
- provide complete Javadoc for the AI module

The AI layer is optional and degrades gracefully when providers
are unavailable or responses fail.

build.gradle

@@ -8,6 +8,10 @@ plugins {
 group = 'org.egothor.methodatlas'
 version = gitVersion(prefix:'release@')
 
+configurations {
+    mockitoAgent
+}
+
 java {
     toolchain {
         languageVersion = JavaLanguageVersion.of(21)
@@ -20,10 +24,27 @@ repositories {
 
 dependencies {
     implementation 'com.github.javaparser:javaparser-core:3.28.0'
+    implementation 'com.fasterxml.jackson.core:jackson-databind:2.21.1'
 
 	testImplementation(platform("org.junit:junit-bom:5.14.2"))
 	testImplementation 'org.junit.jupiter:junit-jupiter'
+	testImplementation 'org.mockito:mockito-core:5.22.0'
+	testImplementation 'org.mockito:mockito-junit-jupiter:5.22.0'
 	testRuntimeOnly 'org.junit.platform:junit-platform-launcher'
+	
+	mockitoAgent('org.mockito:mockito-core:5.22.0') {
+        transitive = false
+    }
+}
+
+tasks.withType(Test).configureEach {
+    useJUnitPlatform()
+    jvmArgs "-javaagent:${configurations.mockitoAgent.singleFile}"
+
+    doFirst {
+        println "Mockito agent: ${configurations.mockitoAgent.singleFile}"
+        println "JVM args: ${jvmArgs}"
+    }
 }
 
 application {