the flakes referenced in #51704 were still seen downstream. the current timeout approach is known to be faulty, but fixing the tests has not been prioritized. this increases the timeout sufficiently to avoid flakes in the meantime