Currently the OplogFetcher checks to see that it hasn't fallen off the oplog by checking if the minTs it requests is in the response. This is OK for unfiltered oplog fetching but kind of messy for filtered oplog fetching (since we must explicitly include that first timestamp, then exclude it according to the filter). We have a query option, ASSERT_MIN_TS_HAS_NOT_FALLEN_OFF_OPLOG, that can be used, we might want to use that instead, either in filtered mode or always.