No app-level timeout on synchronous BigQuery calls — large queries get killed by the worker

# No app-level timeout on synchronous BigQuery calls — large queries get killed by the worker

## Summary
Data-download queries run in-request via `.result().to_dataframe()` with no explicit timeout set, while gunicorn uses its default 30s worker timeout (no `--timeout` flag configured in the Dockerfile). A large/wide query that runs past 30s gets its worker killed mid-request, so the client sees a bare connection reset instead of a clean error response.

## Where this came from
Baalmart, following up on AirQo-frontend PR #3876's data-download review, tracing the request path in `src/analytics`. Flagged as a possible explanation for reported flakiness on repeated large requests.

## Affected code
- Synchronous BigQuery call — `bigquery_api.py:517-524`
- Gunicorn worker timeout (no explicit `--timeout` in the Dockerfile)

## Suggested fix
Set an explicit BigQuery job timeout with a clean error response when exceeded, and tune the gunicorn `--timeout` to give large exports room to complete (or fail cleanly) instead of being killed by the default.

## Acceptance criteria
- [ ] BigQuery data-download queries have an explicit timeout with a clean, client-visible error on expiry
- [ ] Gunicorn worker timeout is tuned to match, so large legitimate queries aren't killed mid-request

Start by reading the synchronous BigQuery call at bigquery_api.py:517-524 and inspect the Dockerfile for the gunicorn command and its missing --timeout setting. Trace how expiry errors are returned to clients, then verify that data-download queries have a client-visible timeout and that the worker timeout allows legitimate large queries to complete or fail cleanly.

Rédigé par le modèle d'indexation à partir du texte de l'issue.