feat(charts/authentik): add probes to worker deployment

[channel-42]

This PR adds probes (liveness, readiness and startup) to the authentik-worker deployment.

Currently, even if a worker pod gets unhealthy (e.g. due to a restart of redis as described here goauthentik/authentik#6221), it does not get restarted by the kubelet due to the missing liveness probe and certain tasks (like adding new providers) do not work, until the affected pod is restarted manually.  

With the proposed changes, in case a worker finds itself in an unhealthy state (returned by ak healthcheck), the kubelet restarts the worker pod and the new healthy worker reconnects to the server pod.

[BeryJu]

I think there was a reason why we don't do this by default anymore, iirc for the worker container, the ak healthcheck has to load the full django app, which uses a lot of ram

[channel-42]

I ran the ak healthcheck command continuously and monitored the memory usage while doing so. No significant increase. Looking through the sources, I found the healthcheck command for the worker, which is just a os.Stat() on the heartbeat file.

[MayurDuduka]

+1

[BeryJu]

normally its inadvisable to have the liveness and readiness probe be the same, but alas we don't have a better probe for celery yet

[channel-42]

The pipeline fails at ct install with client rate limiter Wait returned an error: context deadline exceeded due to the timeout set in .github/configs/ct-install.yaml.

The chart definitely installs fine on my local cluster. Would it be viable to increase the pipeline timeout?