[M3][Epic 3.2] Add Prometheus metrics, structured logging, and correlation IDs

[POWDER-RANGER]

Problem

Implement observability stack per docs/architecture.md requirements.

Tasks

• Add Prometheus client to backend (prom-client)
• Expose /api/metrics endpoint
• Track request counts, durations, and error rates
• Add structured logging (Winston or Pino)
• Generate correlation IDs for each request
• Pass correlation IDs through all service calls (frontend → backend → ML)
• Log correlation ID with every log entry
• (Optional) Add tracing with Jaeger or OpenTelemetry

Acceptance Criteria

• ✅ /api/metrics exposes request counts and latencies
• ✅ Correlation IDs logged end-to-end
• ✅ Can trace a single request across all 3 services

Priority: P1 - High
Labels: observability, infrastructure, M3, P1