Prometheus metrics, /metrics & /health endpoints, DB instrumentation, pool monitor, user-activity middleware, Prometheus rules and Grafana dashboard

[Prashant-thakur77]

What does this PR do?

) This PR adds a minimal, opinionated observability layer to the project so we can collect query performance metrics, monitor connection pools, track user activity, expose system health, and provide Prometheus alerting and a starter Grafana dashboard.

What I changed (summary)

Added a Prometheus-backed metrics collector with histograms/counters/gauges and a simple MetricsCollector class.
app/lib/monitoring/metrics.ts
Exposed metrics and health endpoints:
app/api/metrics.ts
app/api/health.ts
DB instrumentation wrapper to measure query duration and status:
app/lib/db/instrumentedQuery.ts
Connection pool monitor to periodically push pool

Related Issue

Closes #80

How to install & test locally

Install runtime dependency and dev types: npm install prom-client npm install --save-dev @types/express @types/node

Note: @types/prom-client does not exist — do NOT try to install it. Instead either upgrade prom-client (recent versions bundle types) or add a minimal declaration file:

Create types/prom-client.d.ts containing: declare module "prom-client";
Run the TypeScript check: npx tsc --noEmit

Wire into your Express app:

Mount metrics and health routers (e.g., app.use(metricsRouter); app.use(healthRouter);)
Wrap DB calls with instrumentedQuery or patch pool.query
Start pool monitor with a getPoolStats function for your DB pools
Verify:

curl http://localhost/:/metrics → should show mcp_ prefixed metrics
curl http://localhost/:/health → should return basic health JSON
Simulate queries and check histogram/counter updates
Notes and important considerations

Protect /metrics in production (network-only access or strip sensitive labels). High-cardinality labels (raw SQL strings, full user IDs) will explode Prometheus series — prefer canonicalized query names and anonymized user IDs.
If prom-client types are missing, add types/prom-client.d.ts as described or update prom-client to a version that ships types.
I included a sample Prometheus rules file for alerting and a small Grafana dashboard JSON as a starting point — adjust thresholds/labels for your environment.