Description of change
Add in the backend for juju upgrade-controller for k8s controllers.
There's a new caas upgrade worker based off the iaas upgrade worker. The worker simply patches the image patch of the operator image to match the desired version.
Due to a bug in pod restart, the upgrade is kicked off but the pod can't come up again. This is being worked on.
QA steps
bootstrap into k8s upload to docker an operator image tagged with a newer version run upgrade-controller use kubectl to see that the controller pod restarts with the newer version image
ensure upgrade iaas still works
该提问来源于开源项目：juju/juju
