Currently, startup of our controllers fails when the Zalando Operator already exist in the cluster: ``` 2020-12-16T15:53:48.126+0100 ERROR setup unable to install external YAML {"error": "clusterroles.rbac.authorization.k8s.io \"postgres-operator\" already exists"} ``` This would easily be recoverable if we would simply update the existing resource instead of exiting. Also, rolling updates with a newer zalando operator version are not possible without an update?