This test has been flakey for a number of runs on these two bots.

Please triage, and re-assign if needed. (Unfortunately, the regression range flagged by sheriff-o-matic seems unhelpful.)