Simplify the mechanism to extend the timeout of the watchdog handlers.

The current #setOneOffTimeoutForCurrentThread does not work because by setting the status of the checker to complete a new run will be scheduled quickly after with the previous timeout.

Instead we pause the thread for the given time. This is similar to the existing pause/resume mechanism.

Change-Id: I0f2977837342c66f58b3e58ff43ed164680238d5
Bug: 223256297