feat(api): re-resolve corpus_version on demand when empty at startup

fix(api): return 503 with Retry-After on pool exhaustion

refactor(rag): strategy pattern for LLM providers — eliminate provider if-chains

feat(llm): add multi-provider abstraction supporting Groq and OpenAI-compat

feat: add hybrid search (FTS + vector + RRF) to retrieval pipeline

test: add unit and smoke tests for rag pipeline and query endpoint