add restore operation for cluster

when a magnum cluster Upgrade /Resize action failed，the cluster status will be "UPDATE_FAILED" and users can not do a update actions any more. the update_failed status shoud not be the final status for a cluster and in most scenarios we still want to manage the cluster in the future. we know that repair a COE cluster from a failed status is complicated，but before we get a better auto-repair solution，we can manually check the cluster，make sure that it works properly，and offer a special use request to set the cluster status in a restore_complete status, the restore_complete is included in the allow_update_status and then we can redo Upgrade /Resize action in the environment.