Rust ORM Phase 1: Route observability collections through Rust data-daemon worker

[joelteply]

Summary

• Wire the existing Rust data-daemon worker (2,292 lines, Unix socket IPC, per-connection threading) into DataDaemon's per-collection adapter routing
• 10 write-heavy observability collections now route through Rust instead of TypeScript SQLite: cognition_state_snapshots, cognition_plan_records, cognition_plan_step_executions, cognition_self_state_updates, cognition_memory_operations, cognition_plan_replans, adapter_decision_logs, adapter_reasoning_logs, response_generation_logs, tool_execution_logs
• Rust worker's data_create/data_update now filter columns via PRAGMA table_info before INSERT/UPDATE (fixes 57 write errors from TypeScript entities with fields that don't map to SQL columns)
• No fallback: process.exit(1) if Rust worker socket not available at startup; writes fail hard if worker dies mid-operation (proven by killing worker and observing ECONNREFUSED errors)

Verified

• 1,113 Rust worker requests, 779 creates, 0 errors
• Killed Rust worker → writes to routed collections fail with ECONNREFUSED (no silent fallback to TypeScript SQLite)
• Non-routed collections (users, chat_messages) continue through TypeScript SQLite unaffected
• ping, data/list, chat/send all functional

Rollback

Remove await this.connectRustDataWorker(); from DataDaemonServer.initializeDeferred(). One line.

Test plan

• TypeScript + Rust compile clean
• 0 write errors in Rust timing log across 779 creates
• Kill-test: Rust worker death causes write failures (no fallback)
• Non-routed collections still work through TypeScript SQLite
• System ping, chat, data operations all functional