feat: wait for Postgres to be up and running before starting gunicorn

[thelfensdrfer]

Every statement here is a guess of mine, but this worked for me.

The problem for me was, gunicorn started faster than Postgres was ready. So these lines:

tail -Fv /var/log/postgresql/postgresql-16-main.log &
tailpid=${!}

only check if Postgres is running, not accepting new connections. That lead to 404 errors and the api does not recover automatically from a non-running Postgres. See #607 for details. I added a pg_isready loop to wait for Postgres actually being ready before starting the api.