context in https://mongodb.slack.com/archives/C0AEK9FJH8R/p1780482292801419

This can lead to a situation where the primary doesn't produce new oplog entries and secondaries will be out of think until forever. Example: https://spruce.corp.mongodb.com/task/mongodb_mongo_master_enterprise_amazon2023_linux_x8[…]0a_26_06_02_13_26_55/tests?execution=0&sorts=STATUS%3AASC

This could happen in production under cache pressure, if wt issues a rollback/out-of-cache.

crash or restart the requester.