policy for automated pod restart when unresponsive

[joel-bluedata]

Currently if a pod is unresponsive, some other user/client may have to explicitly issue a pod delete to reboot it. This is standard practice for statefulset pods and there are reasons that K8s doesn't just always automatically restart a pod on a new host. However we need to think about whether we should support optional policies for automated pod restart, specifiable in the app CR and/or cluster CR.