Add performance features: caching, cost tracking, retry, compaction, classification, scrubbing

Inspired by zeroclaw's lightweight patterns for slow hardware:
- Response cache (SQLite + SHA-256 keyed) to skip redundant LLM calls
- History compaction — LLM-summarize old messages when history exceeds 50
- Query classifier routes simple/research queries to cheaper models
- Credential scrubbing removes secrets from tool output before sending to LLM
- Cost tracker with daily/monthly budget enforcement (SQLite)
- Resilient provider with retry + exponential backoff + fallback provider
- Approval engine gains session "always allow" and audit log

Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>

pyproject.toml

@@ -19,6 +19,7 @@ dependencies = [
     "json-repair>=0.30.0",
     "duckduckgo-search>=7.0.0",
     "pypdf>=5.0.0",
+    "aiosqlite>=0.20.0",
 ]
 
 [project.scripts]