etcd-restore-operator creates service for itself

[hongchaodeng]

Currently we need to deploy another service for etcd-restore-operator. This adds complexity to deployment.

To further simplify this issue, We can assume etcd-restore-operator is created as a Deployment. We can create the service automatically this this way:

• use the downward API POD_NAME and POD_NAMESPACE to get the pod object.
• get the labels specification
  • but intelligently ignore "pod-template-hash" which is added by deployment
• use the labels to create the service