Add guide for collecting GCP managed service metrics

[thilak009]

Adds instrument/infra/gcp/gcp-cloud-monitoring-to-scout.md, covering the
googlecloudmonitoring receiver for pulling GCP managed service metrics
into Scout.

Why

GCP had exactly one page on the site — the Cloud Logging guide — and no
metrics story at all. That's a gap next to AWS (a five-page
cloudwatch-metrics/ section) and Azure (an overview plus ~22 per-service
pages). This is the metrics companion to the existing logs guide.

What it covers

• The mechanism, and how it differs from logs: the collector polls the
  Cloud Monitoring API, so there is no Log Router sink, Pub/Sub topic,
  subscription, or publisher binding — just roles/monitoring.viewer and a
  metrics list.
• Metric discovery via Metrics Explorer, with deep links to Google's
  exhaustive per-service references (Cloud SQL, Memorystore Redis).
• Authentication via ADC, reusing the service account from the logs guide.
• Full receiver option table (collection_interval defaults to 300s,
  minimum 60s; initial_delay, timeout, endpoint).
• metric_name vs metric_descriptor_filter for collecting a whole
  service by prefix.
• The GAUGE + DISTRIBUTION failure mode, where one bad metric fails the
  entire scrape batch and drops every metric from that receiver instance.
• Troubleshooting, including replica duplication and managed-vs-self-hosted.

Scope

Single GCP project. A cross-project variant is intended as a follow-up,
mirroring the one in the logs guide.

Verification

• markdownlint clean on the new file
• npm run typecheck passes
• npm run build succeeds — onBrokenLinks is throw, so this confirms
  all internal links resolve
• Rendered locally and reviewed

Note: npm run check was not run as-is because its markdownlint -f step
auto-fixes every file in the repo, which would have swept in unrelated
changes. The three underlying checks were run individually instead.

🤖 Generated with Claude Code