Collect CI cost and duration telemetry using the GitHub Actions workflow run API and OpenTelemetry

Verified steps

1. Query completed workflow runs via GET /repos/{owner}/{repo}/actions/runs with status=completed and created=>TIMESTAMP to page through runs since your last collection window; the response includes run_started_at, updated_at, and billable usage per runner OS
2. Drill into job-level timing with GET /repos/{owner}/{repo}/actions/runs/{run_id}/jobs; each job object contains started_at, completed_at, and per-step timing allowing step-level duration analysis
3. Retrieve billable minutes from GET /repos/{owner}/{repo}/actions/runs/{run_id}/timing; the response breaks down UBUNTU, MACOS, and WINDOWS minutes consumed by the run
4. Export metrics to an observability backend using the github-actions-opentelemetry GitHub Action (marketplace action by paper2): it reads completed workflow data via the API and pushes traces and metrics to an OTLP endpoint
5. Alternatively, use the catchpoint/workflow-telemetry-action inside workflows to capture real-time CPU, memory, and I/O metrics from the runner and send them to a configured metrics endpoint during the run
6. Build a dashboard aggregating run duration, queue time (run_started_at minus created_at), and failure rate per workflow name to identify high-cost or flaky pipelines for optimization