P1: Add Prometheus metrics for observability

[BadgerOps]

No Metrics/Observability

Priority: P1 - High
Estimate: 3 hours
Review Reference: REVIEW.md (P1-5)

Issue

No visibility into:

• Request rates and latency
• Error rates
• Resource utilization
• Pool/account counts

Impact: High - Cannot monitor production health

Solution

Add Prometheus metrics with /metrics endpoint.

Key Metrics:

• cloudpam_http_requests_total - Counter with method, path, status labels
• cloudpam_http_duration_seconds - Histogram for latency
• cloudpam_pools_total - Gauge for pool count
• cloudpam_accounts_total - Gauge for account count

Implementation:

1. Add github.com/prometheus/client_golang dependency
2. Create internal/http/metrics.go
3. Instrument key operations
4. Add /metrics endpoint in RegisterRoutes()

See REVIEW.md (P1-5) for complete code.

Acceptance Criteria

• Prometheus client_golang dependency added
• metrics.go created with key metrics
• HTTP requests instrumented with counters and histograms
• Pool/account counts exposed as gauges
• /metrics endpoint returns Prometheus format
• Tests verify metric collection
• CLAUDE.md updated with metrics documentation

Related

• Consider OpenTelemetry tracing (optional) #41 (OpenTelemetry) for distributed tracing
• Complements Rate limiting and structured request logs #17 (structured logging)