Owls 93071 - Display correct readyReplicas and server state when domain serverStartPolicy switched to NEVER

[ankedia]

The incorrect readyReplicas was because the status update fiber was using a stale DPI cache for the list of ready pods. With this change, the StatusUpdate step that runs periodically makes use of the DPI cached in the DomainProcessorImpl instead of DPI passed when scheduling the domain status updating.

The change also deletes the server pod from DPI when the pod is not found during a list operation and is fully deleted. The server state is displayed as SHUTTING_DOWN when the pod is being deleted and SHUTDOWN after the pod is deleted.

Added unit test to verify the changed behavior. The integration test results are at - https://build.weblogick8s.org:8443/job/weblogic-kubernetes-operator-kind-new/6762/

[rjeberhard]

@ankedia, I believe that you told me that this change would not need to be forward ported to main. It seems like this change covers a lot of edge cases. I would want to be sure that none of these edge cases still need to be covered on main.

[ankedia]

@ankedia, I believe that you told me that this change would not need to be forward ported to main. It seems like this change covers a lot of edge cases. I would want to be sure that none of these edge cases still need to be covered on main.
@rjeberhard - ok, I will run the tests against the main branch to see if same issue exists in main and work on forward porting these changes accordingly. Since lot of domain status functionality has changed in the main branch, I'm not sure if same changes would apply or if they need to be done differently.