Ln 651 Query Efficiency Auditor

Audits database query patterns for common performance killers: redundant fetches where the same entity is loaded twice in a call chain, N+1 update loops that should be batch operations, over-fetching full models when you only need three fields, and request-scoped caches for data that changes once a week. Uses a two-layer detection system to avoid false positives and supports hex-graph MCP tools to trace call paths when available. Returns findings with severity ratings, effort estimates, and a compliance score. Most useful during code review or when you're trying to figure out why your API suddenly needs 47 database round trips per request.