feat(observability): Grafana dashboards for API and pipeline health

[chitrank2050]

Problem

Metrics without dashboards are data without insight. Operators need visual dashboards to monitor system health at a glance.

Proposed solution

Create Grafana dashboard JSON files (provisioned via docker-compose):

1. API Golden Signals: request rate, error rate, latency p50/p95/p99, active connections
2. Usage Pipeline: events ingested/sec, validation rate, aggregation rate, DLQ growth, outbox backlog
3. Billing Cycle: invoices generated/day, payment success rate, average invoice amount, overdue count
4. Infrastructure: Postgres connections, Redis memory, Kafka consumer lag, Node.js event loop lag

Each dashboard has: time range selector, auto-refresh, tenant filter variable.

API design (if applicable)

N/A — Grafana dashboards provisioned via JSON files.

Alternatives considered

• Manual dashboard creation — rejected. JSON provisioning is reproducible and version-controlled.
• Datadog dashboards — rejected. Grafana is open-source and already in our Docker stack.

Additional context

Add Grafana + Prometheus to docker-compose.yml. Grafana at http://localhost:3001. Dashboards auto-loaded from grafana/dashboards/.