Request serving scheduler not cleaning up old placeholder deployments

  When using cluster size tagging and related request serving node scheduler, if a cluster is deleted while in the middle of resizing its request serving pods, the placeholder deployment that was created for it is not cleaned up.

   HyperShift main (4.16)

   Always 

    1. Setup a management cluster with request-serving machinesets
    2. Create a hosted cluster
    3. Add workers to the hosted cluster so that it changes size
    4. Delete the hosted cluster after the it's tagged with the new size but before nodes for its corresponding placeholder pods are created.
    

    The placeholder deployment is never removed from the `hypershift-request-serving-autosizing-placeholder` namespace

    The placeholder deployment is removed when the cluster is deleted.