Observability bundle: query log, health checks, structured logging

[eaguad1337]

Problem

There is no database query profiling, no health endpoints, and no structured logging that correlates the lines emitted during a single request. This makes debugging and production operations harder.

Proposal

An observability bundle (may be split into sub-issues):

• Query log / profiler — capture query count, per-query timing, and a slow-query flag, exposed as a data source the debugbar can render.
• Health checks — a /health endpoint (liveness + readiness) backed by a registry of pluggable dependency checks (database, cache, queue, storage).
• Structured logging — a JSON log formatter and a per-request correlation id propagated through the logger.

Scope

ORM query hooks, debugbar data source, health-check registry + endpoint, JSON formatter + correlation middleware.

Definition of Done

• Implementation (sub-issues may track each piece).
• pytest covers the query recorder, health registry, and JSON/correlation output (suite green).
• Docs page(s) + nav entry.
• Release notes entry.