Resolve "Restarted pods sometimes fail to init"

Closes #56 (closed)

I always thought this was caused by a problem in the storage layer, but it turned out that a part of the ansible playbook changed the rights to make the config file read-only. It also tried to change the user to www-data, which does not have the same user ID in the init container as the PHP container. For that reason I removed the www-data stuff from values.yaml (it wasn't used consistently any way). I re-set the config file to writable just before we try to edit it. The part that makes it read-only still exists, because it kinda makes sense security-wise.