If an app's jobs keep timing out, then the app potentially has a bug.
Timeouts resulted in backoff of the individual job that timed out,
but didn't affect other jobs. This meant that apps could bypass backoff
by scheduling one-off jobs instead of rescheduling using the return
parameter to onStopJob. Now, JobScheduler will track the number of job
timeouts across all of the app's jobs and reduce the app's max job time
if it exceeds a threshold. If the total number of timeouts exceeds
another limit, JobScheduler will consider the app buggy and
attempt to put the app in the RESTRICTED bucket.

Similarly, if an app keeps ANRing from JobScheduler's perspective, it
will consider the app buggy and attempt to put it in the RESTRICTED
bucket.

When an unexempted app is considered buggy, don't treat any of its jobs
as in the ACTIVE unless the app is current active.

Also, reduce the delay for forcing an app in the RESTRICTED bucket
(eg. when the app is buggy) to 1 hour. The normal bucket timeout
continues to be 8 days.

Bug: 284080732
Test: atest AppStandbyControllerTests
Test: atest FrameworksMockingServicesTests:JobSchedulerServiceTest
Test: atest FrameworksMockingServicesTests:JobStatusTest
Test: atest frameworks/base/services/tests/mockingservicestests/src/com/android/server/job
Test: atest frameworks/base/services/tests/servicestests/src/com/android/server/job
Change-Id: I8eb333a922658ba6137012dffd7a143cd990376d