Add workspace and memory system (OpenClaw-inspired)

Implements persistent memory for agents with hybrid search:

- Database-backed workspace with PostgreSQL (not filesystem)
- Memory documents: MEMORY.md, daily logs, identity files
- Chunked content with FTS (tsvector) + vector (pgvector) indexes
- Reciprocal Rank Fusion (RRF) for hybrid search combining BM25 and semantic
- Memory tools: memory_search, memory_write, memory_read
- Proactive heartbeat system for periodic execution (30 min default)
- OpenAI embeddings provider (text-embedding-3-small)

Key patterns from OpenClaw:
- "Memory is files, not RAM" - explicit persistence required
- Two-tier memory: daily logs (raw) + curated MEMORY.md
- Session isolation via user_id/agent_id scoping

Co-Authored-By: Claude Opus 4.5 <[email protected]>

Cargo.lock

@@ -1275,6 +1275,7 @@ dependencies = [
  "deadpool-postgres",
  "dotenvy",
  "futures",
+ "pgvector",
  "postgres-types",
  "pretty_assertions",
  "refinery",
@@ -1404,6 +1405,16 @@ version = "2.3.2"
 source = "registry+https://github.com/rust-lang/crates.io-index"
 checksum = "9b4f627cb1b25917193a259e49bdad08f671f8d9708acfd5fe0a8c1455d87220"
 
+[[package]]
+name = "pgvector"
+version = "0.4.1"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "fc58e2d255979a31caa7cabfa7aac654af0354220719ab7a68520ae7a91e8c0b"
+dependencies = [
+ "bytes",
+ "postgres-types",
+]
+
 [[package]]
 name = "phf"
 version = "0.13.1"