Polish: BGE-large embeddings, contextual retrieval, 142 tests passing, lint clean

Security, bug fixes, and UX improvements

Add A-RAG — hierarchical retrieval interfaces (Feb 2026)

Add Speculative RAG — Google Research 2024 (~51% latency reduction)

Add TTRAG — test-time compute scaling for RAG (ICLR 2025)

Initial commit — production RAG system