Observability, telemetry, and operational dashboards

## Summary
Implement observability, telemetry, and operational dashboards for self-hosted production deployments.

## Scope
- Instrument key product flows such as auth, LiveView joins, message send, media handling, moderation, and background jobs.
- Provide operator-facing dashboards or documented metrics surfaces.
- Define actionable signals for latency, error rate, queue depth, and realtime health.
- Support debugging of production incidents without attaching a debugger to the live system.

## Acceptance Criteria
- [ ] Critical product flows emit metrics and structured telemetry.
- [ ] Operators can inspect realtime health, error rates, and resource pressure from a supported interface.
- [ ] Alert-worthy failure classes are identifiable from exported metrics or logs.
- [ ] Instrumentation adds clarity without materially distorting performance.

## Operational Notes
- Self-hosted operators vary widely in sophistication, so the defaults must be useful without a large observability stack.
- Benchmarks and load work should map onto the same telemetry vocabulary.