feat(observability): enhanced health check with dependency latency

[chitrank2050]

Problem

Current health check only verifies Postgres connectivity. Redis and Kafka status are unknown. No latency information — a 10-second database response still shows "healthy."

Proposed solution

Enhance GET /health to report all dependencies:

{
  "status": "ok",
  "dependencies": {
    "postgres": { "status": "up", "latencyMs": 2 },
    "redis": { "status": "up", "latencyMs": 1 },
    "kafka": { "status": "up", "latencyMs": 5 }
  },
  "uptime": 86400,
  "version": "0.6.0"
}

Each dependency check: ping/NOOP command, measure round-trip, timeout at 5 seconds.

Alternatives considered

• Separate liveness and readiness probes — future enhancement for Kubernetes. Start with unified /health.

Additional context

Uses @nestjs/terminus which is already installed. Add Redis and Kafka health indicators.