Remove query rewrite step from RAG search pipeline

v1.7: parent chunk injection, atomic facts, MCP server, gpt-5-nano pipeline

Dual timestamps, query decomposition, text-embedding-3-large

Add PostgreSQL + pgvector backend for production use

Async I/O, read/write locks, v1 architecture doc

Rewrite README in first-person narrative style

RAG: 82.8% accuracy, disk persistence, K=20 retrieval

RAG provider: 80.8% on LongMemEval (500q), 5.1 points behind supermemory

Merge pull request #25 from supermemoryai/claude/add-memory-tools-HgAI7

supermemory hybrid mode change

Add LLM memory extraction to filesystem and RAG providers

Add filesystem and RAG memory providers for benchmarking

Merge pull request #17 from supermemoryai/temporal-date-bug

[bug] temporal date fallback

Merge pull request #14 from supermemoryai/feat/parallel-execution