Migrate jobs to postStartCommand

As Maarten suggested, we might want to replace the setup-apps job and the configure-nextcloud one by postStart commands.

The upstream nextcloud helm chart supports adding custom extra volumeMounts to inject the scripts into the nextcloud pod.

Disadvantages

• Makes nextcloud start-up slower.
• The postStart event gets sent early, when the container has just been created. Nextcloud is not yet up then, so we'll have wait and poll until it's available. Also, we may want to build in a simple retry mechanism for when nextcloud (or mariadb, or redis) has a temporary failure. These are things that kubernetes does for us currently with the jobs mechanism.

Advantages

• Simplifies the chart, because we no longer have to keep these separate jobs configured correctly, duplicating things like volume mounts and environment variables. Failing to keep that duplication in order may have been the cause of #1462.