Watcher Pods Sticking Around

[gambol99]

Describe the bug
The namespace pods for the configuration watcher seem to stick around for quite a while. This could be delated to 'Completed' vs 'Error', not sure - but we need to look into it.

Expected behavior
When the job is removed - all pods are removed from it.