During migrations with MTC the new registry pod dies and migration never completes

Issue

The registry pod is frequently removed and recreated:

$ oc get pods
NAME                                                              READY   STATUS    RESTARTS   AGE
migration-log-reader-74bfbd6cb6-pldpm                             2/2     Running   1          27h
migration-operator-5fd68c7c65-jfpfd                               1/1     Running   1          27h
registry-668e4c06-8f6d-4c8f-b34d-e03845428e67-ng7fx-74b94cctd5h   1/1     Running   0          5m
restic-2rmlh                                                      1/1     Running   1          27h
...

This operation is also reflected in the project events:

$ oc get events
LAST SEEN   TYPE     REASON              OBJECT                                                                      MESSAGE
19m         Normal   SuccessfulCreate    replicaset/registry-668e4c06-8f6d-4c8f-b34d-e03845428e67-bmqlv-6cb8c879b9   Created pod: registry-668e4c06-8f6d-4c8f-b34d-e03845428e67-bmqlv-6cb8c8nd6pf
19m         Normal   Scheduled           pod/registry-668e4c06-8f6d-4c8f-b34d-e03845428e67-bmqlv-6cb8c8nd6pf         Successfully assigned openshift-migration/registry-668e4c06-8f6d-4c8f-b34d-e03845428e67-bmqlv-6cb8c8nd6pf to worker4-lab
19m         Normal   Pulled              pod/registry-668e4c06-8f6d-4c8f-b34d-e03845428e67-bmqlv-6cb8c8nd6pf         Container image "registry.redhat.io/rhmtc/openshift-migration-registry-rhel8:v1.7.4-7" already present on machine
19m         Normal   Created             pod/registry-668e4c06-8f6d-4c8f-b34d-e03845428e67-bmqlv-6cb8c8nd6pf         Created container
19m         Normal   Started             pod/registry-668e4c06-8f6d-4c8f-b34d-e03845428e67-bmqlv-6cb8c8nd6pf         Started container
13m         Normal   Killing             pod/registry-668e4c06-8f6d-4c8f-b34d-e03845428e67-bmqlv-6cb8c8nd6pf         Killing container with id docker://registry:Need to kill Pod
19m         Normal   ScalingReplicaSet   deployment/registry-668e4c06-8f6d-4c8f-b34d-e03845428e67-bmqlv              Scaled up replica set registry-668e4c06-8f6d-4c8f-b34d-e03845428e67-bmqlv-6cb8c879b9 to 1
5m          Normal   SuccessfulCreate    replicaset/registry-668e4c06-8f6d-4c8f-b34d-e03845428e67-ng7fx-74b94cbd55   Created pod: registry-668e4c06-8f6d-4c8f-b34d-e03845428e67-ng7fx-74b94cctd5h
...

Registry pod can end in crashLoopBackOff status and migration is blocked as conditions are not meet:

Conditions:
    Category:              Critical
    Last Transition Time:  2022-11-14T10:56:12Z
    Message:               Reconcile failed: [unable to retrieve the complete list of server APIs: metrics.k8s.io/v1beta1: the server is currently unable to handle the request]. See controller logs for details.
    Status:                True
    Type:                  ReconcileFailed
Events:
  Type     Reason           Age                    From                    Message
  ----     ------           ----                   ----                    -------
  Warning  Postponed        6m32s (x2 over 6m42s)  miganalytic_controller  Waiting 10 seconds for referenced MigPlan to become ready.
  Warning  ReconcileFailed  95s (x29 over 6m21s)   miganalytic_controller  Reconcile failed: []. See controller logs for details.