Was previously not handled at all, therefore displaying request failures
in the test log output. While that was mostly a red herring and
shouldn't have caused any actual *test* failures, it is still better to
handle this explicitly in a realistic manner.
Because submission can be prevented by both not having been issued a
correct submission token, and by not actually registering any hits in
gameplay, ensure that tests that don't receive a token register at least
one hit, to avoid potentially having test cases that test the "no token"
flow pass erroneously because they never had any hits in the first
place.
