Description
This stat in the UI is derived from the underlying stat docs_processed.
docs_processed correctly includes docs which have been checked but did not need to be replicated as the winning revision is already on the destination cluster.
If, for example, the same data was loaded independently on two clusters, and then a replication set up, we would still report that mutations had been replicated, even though no data had been sent from the source to destination cluster. This is quite misleading, perhaps it would be best to re-label the UI stat?