Docs: subsystem honesty pass + CORTEX ecosystem map

[Rezzyman]

This release brings the README into accurate alignment with what cortex-python actually implements today, plus adds the shared ecosystem reference.

Subsystem parity update

The architecture section now explicitly lists which subsystems are implemented in this Python release and which are part of full CORTEX (TypeScript) but not yet ported:

Implemented: ingestion pipeline, hippocampal encoding (DG + CA1), dream cycle, procedural memory, 5-factor hybrid search, temporal validity, CLI.

Not yet in this release: CA3 pattern completion, reconsolidation/labile window, emotional valence (6-dimensional), metacognition, proprioception, autonomous cognition, MCP server, social graph.

Benchmark attribution clarified

The LongMemEval 500/500 and LoCoMo 93.6% scores are correctly attributed to the TypeScript flagship. cortex-python reads and writes the same memory store (wire-compatible Postgres+pgvector schema) but uses 5-factor hybrid search (vs 7-factor in the flagship), so its retrieval has not been independently benchmark-certified.

Search scoring corrected

README now accurately says "5-factor hybrid search" instead of the previously claimed "7-factor" (cortex-python does not yet implement the emotional_boost and ca3_activation scoring factors).

Ecosystem section added

Same three-project comparison table as the other repos, with clear audience framing for when to use cortex-python vs cortex vs cortex-lite.

Project	Language	Storage	Audience
cortex	TypeScript	PostgreSQL + pgvector	Production teams, benchmark-certified retrieval
cortex-lite	Python	SQLite	Individual devs, zero-config, 30-second quickstart
cortex-python	Python	PostgreSQL + pgvector	Python codebases sharing a CORTEX memory store