Currently, tenant oplog applier doesn't commit the no-op writes, but still forwarding the _lastAppliedOpTimesUpToLastBatch's recipient opTime. This can make the donor opTime waiters (recipientSyncData cmd) to hang indefinitely waiting on a recipient opTime that doesn't correspond to any oplog entry to be majority replicated.
+ Change streams are also affected as it won't be able to see the progress.