Currently, the ReshardingOplogFetcher only updates the exponential moving average time to fetch oplog entries upon fetching a batch with a postBatchResumeToken. SERVER-106550 made ShardRemote::runAggregation go back to only returning postBatchResumeToken when the batch is empty. This essentially made the ReshardingOplogFetcher only update the average when it happens to get back an empty because there are no remaining oplog entries that the recipient needs to copy from that donor, i.e. when it has reached the end of the oplog on the node it is fetching from. This may not happen very often if there is a high write rate to the collection being resharded.