docs: Update README to reflect 6 collections (3 RAG + 3 Memory)
Architecture clarification:
- Updated schema section: 4 → 6 collections
- Clarified separation: RAG (3) vs Memory (3)
- Removed Document collection references
- Updated collection names: Chunk → Chunk_v2, Summary → Summary_v2
Schema changes reflected:
- RAG: Work, Chunk_v2, Summary_v2 (schema.py)
- Memory: Conversation, Message, Thought (memory/schemas/memory_schemas.py)
Vectorization details:
- All 5 vectorized collections use GPU embedder (BAAI/bge-m3, RTX 4070)
- Manual vectorization with Python PyTorch CUDA
- 1024 dimensions, cosine similarity
Updated diagrams:
- Architecture mermaid diagram shows 6 collections
- Pipeline diagram updated to 6 collections
- Added memory/ module structure
Updated examples:
- Replaced Chunk with Chunk_v2 in all code examples
- Added Memory collections documentation
- Clarified separation of concerns
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>