Currently once you deploy your application, the operator will never update the runtime image version, even if the app specs are updated. This is done on purpose, to ensure consistency overtime of an application.

With the new stateful applications, it might be tricky to delete and recreate an application because the disk is not retained. It becomes hard to upgrade the runtime image version without impacting the application itself.

One possible solution would be to add a flag in the update command to also refresh the new docker image (if the operator has a new one configured). In that case, the image would be modified and all the pods restarted