Production Deployment: Staging → Production Cutover, Monitoring, Alerting

[snowfox1003]

Problem

The GCP staging deployment was completed in Week 20 (Docker image build, Cloud SQL configuration, deploy-and-validate workflow). The system now runs successfully on staging but has no production environment, meaning all collected community data—from GitHub, Slack, Discord, mailing lists, YouTube, and WG21—remains inaccessible for real reporting. Without production deployment and production-grade monitoring, the platform cannot serve its purpose as a live analytics pipeline, and silent data gaps would go undetected.

Acceptance Criteria

• Production GCP environment provisioned: Cloud SQL (PostgreSQL 16), Cloud Run or GCE instances for gunicorn + Celery worker + Celery Beat + Redis
• Production deployment pipeline (.github/workflows/deploy.yml) updated with production environment-scoped GitHub Secrets and production target
• All 15 apps' migrations applied cleanly to production Cloud SQL; smoke test confirms health endpoint responds 200
• Celery Beat schedule loads boost_collector_schedule.yaml and at least one full collection cycle (e.g., daily/github group) completes without error
• Cloud Monitoring (or equivalent) alerting configured for: container restarts, Celery worker heartbeat loss, collection task failures, and database connection errors
• Structured logging (JSON) enabled so log-based alerting can detect the silent group-exit failure mode identified in the review eval
• Staging environment kept operational as a pre-production gate