A few jstests assert that after a timeseries write, there was exactly one attempt to reopen a bucket. In suites with "non-canonical buckets" (e.g. with chunk migrations) that check is relaxed to "one or more" due to retries.

However due to SERVER-122949, during viewless timeseries upgrade/downgrade we may not find a bucket to reopen. For this reason we have relaxed the check to just a sanity check that the count doesn't go down.

After viewless timeseries is released we can strengthen those checks again.

See TODOs associated to this ticket for affected tests.