TRR BACKEND: Operational visibility for pipeline runs

[therealityreport]

Goal: Make pipeline runs easier to debug without digging through raw DB rows.

TODO (pick subset)

• Add CLI command: python -m trr_backend.cli pipeline status --verbose <run-id> (stage timing + error_details)
• Add --json flag for machine-readable output (future automation)
• Stage timings summary output:
  • stage name, status, duration_ms, processed/skipped/failed, error_message (if any)
• Add CLI command: pipeline logs <run-id> (if logs are stored somewhere)
• Include manifest_key in status output + optional presigned URL generator (service role only, if AWS tooling exists)
• Add pipeline.run_events table (optional) for structured events:
  • stage started/finished
  • counts
  • warnings
• Add S3 manifest enrichment:
  • include tool versions / git SHA / runtime metadata
• Add optional Sentry breadcrumbs (if Sentry is configured)

Acceptance

• Operator can run one command and know which stage failed and why, without opening Supabase UI
• An operator can identify "what broke and where" within 60 seconds using CLI + DB