The new cloners don't (and can't) use the NetworkInterfaceMock, and so don't run in lockstep with the unit test code. This results in the timing sometimes not being as the tests expect. We can use the cloner failpoints to synchronize the tests with the code under test.