I got the debug output of test_3_join_in_forked_from_thread() using some hacks (threading._VERBOSE=True and don't hide subprocess output):

Testing with flags: sys.flags(debug=0, inspect=0, interactive=0, optimize=0, dont_write_bytecode=0, no_user_site=0, no_site=0, ignore_environment=0, verbose=0, bytes_warning=0, quiet=0)
[  1] test_threading
test_3_join_in_forked_from_thread (test.test_threading.ThreadJoinOnShutdown) ... <OS thread 139683761760000>: <Condition(<_thread.lock object at 0x7f0aa24b8e50>, 0)>.notify(): no waiters
MainThread: <Thread(Thread-1, initial)>.start(): starting thread
<OS thread 139683646777088>: <Condition(<_thread.lock object at 0x7f0aa24b8b80>, 1)>.notify(): notifying 1 waiter
Thread-1: <Thread(Thread-1, started 139683646777088)>._bootstrap(): thread started
MainThread: <Condition(<_thread.lock object at 0x7f0aa24b8b80>, 0)>.wait(): got it
MainThread: <Condition(<_thread.lock object at 0x7f0aa24b8cc0>, 0)>.notify(): no waiters
MainThread: <_MainThread(MainThread, stopped 139683761760000)>: waiting for other threads
MainThread: <Thread(Thread-1, started 139683646777088)>.join(): waiting until thread stops
end of main
Thread-1: <Thread(Thread-2, initial)>.start(): starting thread
<OS thread 139683638384384>: <Condition(<_thread.lock object at 0x7f0aa24b8b30>, 1)>.notify(): notifying 1 waiter
Thread-2: <Thread(Thread-2, started 139683638384384)>._bootstrap(): thread started
Thread-1: <Condition(<_thread.lock object at 0x7f0aa24b8b30>, 0)>.wait(): got it
Thread-2: <_MainThread(MainThread, stopped 139683761760000)>.join(): thread stopped
Thread-1: <Thread(Thread-2, started 139683638384384)>.join(): waiting until thread stops
end of thread
Thread-2: <Thread(Thread-2, started 139683638384384)>._bootstrap(): normal return
Thread-2: <Condition(<_thread.lock object at 0x7f0aa24b8b80>, 1)>.notify(): notifying 1 waiter
Thread-1: <Condition(<_thread.lock object at 0x7f0aa24b8b80>, 0)>.wait(): got it
Thread-1: <Thread(Thread-2, stopped 139683638384384)>.join(): thread stopped
Thread-1: <Thread(Thread-1, started 139683646777088)>._bootstrap(): normal return
Thread-1: <_RLock owner='Thread-1' count=1>.acquire(True): initial success
Thread-1: <Condition(<_RLock owner='Thread-1' count=1>, 0)>.notify(): no waiters
Thread-1: <_RLock owner=None count=0>.release(): final release
Thread-1: <Thread(Thread-1, started 139683646777088)>._bootstrap(): raised SystemExit
Thread-1: <Condition(<_thread.lock object at 0x7f0aa24b8bd0>, 1)>.notify(): notifying 1 waiter
MainThread: <Condition(<_thread.lock object at 0x7f0aa24b8bd0>, 0)>.wait(): got it
MainThread: <Thread(Thread-1, stopped 139683646777088)>.join(): thread stopped
MainThread: <_MainThread(MainThread, stopped 139683761760000)>: exiting
[47121 refs]
ok

----------------------------------------------------------------------
Ran 1 test in 0.095s

OK
[  2] test_threading
test_3_join_in_forked_from_thread (test.test_threading.ThreadJoinOnShutdown) ... <OS thread 140303874844416>: <Condition(<_thread.lock object at 0x7f9b03ea4e50>, 0)>.notify(): no waiters
MainThread: <Thread(Thread-1, initial)>.start(): starting thread
<OS thread 140303759861504>: <Condition(<_thread.lock object at 0x7f9b03ea4b80>, 1)>.notify(): notifying 1 waiter
Thread-1: <Thread(Thread-1, started 140303759861504)>._bootstrap(): thread started
MainThread: <Condition(<_thread.lock object at 0x7f9b03ea4b80>, 0)>.wait(): got it
MainThread: <Condition(<_thread.lock object at 0x7f9b03ea4cc0>, 0)>.notify(): no waiters
MainThread: <_MainThread(MainThread, stopped 140303874844416)>: waiting for other threads
MainThread: <Thread(Thread-1, started 140303759861504)>.join(): waiting until thread stops
end of main