Description of problem

The image-registry COS instance created by IPI can fail to delete during the destroy cluster command. It has to be deleted manually in order to re-run the destroy cluster command successfully to complete cleanup.

Version-Release number of selected component (if applicable):
4.12

How reproducible:
Infrequent

Steps to Reproduce:
1. Create a new IPI cluster on IBM Cloud
2. Destroy the IPI cluster

Actual results:
destroy cluster fails attempting to delete the image-registry COS instance

Expected results:
Successful destroy cluster and all IBM Cloud resources have been deleted

Additional info:
I will open tracking issues for IBM Cloud to investigate the cause further, since it appears the calls being made to IBM Cloud API's result in failure, but attempts to manually delete, say in the IBM Cloud UI, appear to work just fine.

Given this is infrequent, my assumption is an issue with the underlying components in IBM Cloud and COS, which hopefully those teams can share some insight into.