The batched_multi_deletes_failover.js might get hung if all step ups fail. This test relies on the fact that a step down aborts the operation context, which in turn cancels a sleep done by the failpoint on which this test relies. If for some reason all step ups fail and thus there is no step down, the sleep within the failpoint is never cancelled and the test gets stuck due to the huge value for sleep (1 hour) until evergreen kills the task.

Let's also lower the number of documents in the collection so as to prevent the batchedDeleteStageSleepAfterNDocuments from failpoint from timing out on slower instances, like on debug builds.