[Epic] Observability & Monitoring

## Summary

The Canton Middleware has a foundational monitoring setup — Prometheus metrics are defined, health endpoints exist on all three services, and structured logging (Zap) is in place — but the implementation is incomplete and not operationally useful. Most metrics are never emitted, there are no dashboards, no alerting rules, no tracing, and Prometheus itself isn't part of the local dev stack.

This tracker covers the end-to-end work to bring the middleware to production-grade observability.