[ai-core] feat: langchain4j-based local mock implementations

[Schmarvinius]

Scope: enable token-free local & CI testing.

• Replace MockAICoreServiceImpl.inferenceClient UnsupportedOperationException with a langchain4j-backed ApiClient for local models.
• Provide a deterministic MockRecommendationClient (drop Random) optionally backed by langchain4j; add a class-level javadoc explaining the prediction algorithm.
• Externalize model mapping in application.yaml (sketch from agoerler):
  cds.ai.core.client:
  • name: myRpt1Client
    model: RPT1
    → enables @qualifier("myRpt1Client") ApiClient injection and cheap models in test, expensive in prod.
• Document local-model usage in both READMEs.

[github-actions]

👋 Hello @Schmarvinius, thank you for submitting this issue. Our team is reviewing your report and will follow up with you as soon as possible.

[Schmarvinius]

Additional review feedback from PR #49

• ApiClient is low-level: On top of ApiClient we could (later) have integrations with the ai-sdk-java and other higher-level APIs like LangChain4j.
• Integration point: Do we want to have an integration point for ai-sdk-java and LangChain4j? How would this look like?