Healthcheck of celery worker does not account on the health of the fastapi-app that spawns on a thread

[pcrespov]

We found the following situation:

• the fastapi-app inside staging-simcore_staging_sto-worker was failing the on_startup event to connect to postgres constantly
• the swarm manager observes that staging-simcore_staging_sto-worker is healthy so it does not restart it
• the service staging-simcore_staging_sto-worker is in reality not functional since it cannot connect to the database

Suggestions:

• A mechanism that reflects the health of the spawned fastapi-app in the staging-simcore_staging_sto-worker healthcheck