feat(lab8): Prometheus+Grafana golden signals + error-rate alert

[blacktree-lab]

Goal

Add Prometheus + Grafana to the Lab 6 Compose stack, provision a four-golden-signal dashboard for QuickNotes, and define one good symptom alert (error ratio > 5% sustained 5 min) with a runbook.

Changes

• Added monitoring/prometheus/prometheus.yml - scrapes QuickNotes (quicknotes:8080) every 15s
• Added monitoring/prometheus/alerts.yml - HighErrorRate alert: 4xx+5xx ratio > 5% for: 5m, severity: page, runbook annotation
• Added monitoring/grafana/provisioning/ - auto-provisions the Prometheus datasource + a 4-panel dashboard (Traffic, Errors, Latency-proxy, Saturation)
• Extended compose.yaml - prometheus (:9090) and grafana (:3000) services, pinned images (prom/prometheus:v3.1.0, grafana/grafana:13.0.3), depends_on: service_healthy, no default Grafana creds
• Added docs/runbook/high-error-rate.md - what it means, triage, mitigations, post-incident
• Added submissions/lab8.md + screenshots

Testing

• curl .../api/v1/targets | jq '.data.activeTargets[].health' -> "up" (Prometheus scraping QuickNotes)
• Grafana auto-loads "QuickNotes - Golden Signals" with all 4 panels populated after ~200 mixed requests (screenshot)
• Under sustained ~20% error injection, HighErrorRate crosses 5% and enters pending (18.15% value, severity: page, annotations resolving) - captured; a single 4xx burst does not fire it (sustained-breach gate)

Checklist

• Title is a clear sentence (<= 70 chars)
• monitoring/ with prometheus config, alert rule, and grafana provisioning
• Prometheus scrapes QuickNotes (up == 1)
• Grafana dashboard provisioned with 4 golden-signal panels
• docs/runbook/high-error-rate.md complete (4 sections)
• submissions/lab8.md covers Task 1 + Task 2 with design answers a-g
• Alert observed in FIRING state
• Commits are signed