[DCV-3942] Document DAG parse times via the Airflow REST API

[gams87]

Adds a "Find Slow-Parsing DAGs with the REST API" how-to under How-tos > Airflow, covering how to get per-DAG parse durations for Team Airflow on Airflow 3 environments.

On Airflow 3 the dag-processor records last_parse_duration / last_parsed_time per DAG and GET /api/v2/dags exposes both, so users can get the real parsing-time report (the numbers that matter for scheduling latency, unlike running airflow dags report on a worker) with a plain API call -- no dag-processor pod access, no extra RBAC.

The guide's primary auth path is the environment's Airflow API Token from the launchpad (Environments > Keys), used directly as a Bearer token against the Airflow API URL shown on that same page. An "authenticate as your own user" section covers the alternative flow for long-lived automation and multi-environment use: exchanging a personal Datacoves API key for a short-lived JWT at POST /auth/token via the __datacoves_token__ sentinel username (per-user permissions, non-expiring key, works against any environment the user can access).

Includes the paginated Python script that prints DAGs slowest-first (same shape as the datacoves my parse-logs report), guidance on interpreting parse times, and the Airflow 2 limitation note (its API does not expose parse durations). My Airflow users are pointed at the existing datacoves my parse-logs command instead, which is why the page lives under Team Airflow how-tos.

Verified end-to-end on a live Airflow 3 environment: launchpad token + Bearer against /api/v2/dags returned the full slowest-first report (24 DAGs), and the sentinel exchange flow was verified against the auth manager code and a real cluster's routing (public /api/ and /auth/ bypass pomerium).

🤖 Generated with Claude Code