Description of problem:
The kube-storage-version-migrator migrator deployment is unschedulable if there are no worker nodes available, causing the cluster operator to degrade.

This can cause problems in managed OpenShift environments where all control plane components should be able to be scheduled even if the worker pool is scaled to zero.

Version-Release number of selected component (if applicable):


How reproducible:
Consistent


Steps to Reproduce:
1. Scale worker pool to zero

Actual results:
The migrator workload pod is in a Pending state even thought the operator pod itself will schedule onto the control plane nodes. The cluster operator will degrade.

Expected results:
The migrator pod would reschedule onto the control plane nodes.


Additional info: