The image prepulling pod calls docker directly to pull images. If the pod
hasn't finished before running the resource usage tracking test, there'd be a
cpu spike in docker. We'd rather wait and fail if this is the case, before
running the test.