Don't update lastTransitionTime unless condition status changes

Fixed by @wking, opening bug for Jira linking.

The cluster-dns-operator sets the status condition's lastTransitionTime whenever the status (true, false, unknown), reason, or message changed on a condition.  

It should only set the lastTransitionTime if the condition status changes. Otherwise this can have an affect on status flapping between true and false.  See https://github.com/openshift/api/blob/master/config/v1/types_cluster_operator.go#L129

4.15 and earlier

100%

1. Put cluster-dns-operator in a Degraded condition by stopping a pod, notice the lastTransitionTime
2. Wait 1 second and stop another pod, which only updates the condition message

Notice the lastTransitionTime for the Degraded condition changes when the message changes, even though the status is still Degraded=true

The lastTransitionTime should not change unless the Degraded status changes, not the message or reason.