If a downgrade previously failed and the cluster is still in a transitionary FCV state, we will not allow upgrading the cluster if cleaning up internal metadata was started but not completed.

Description of Linked Ticket

Same as SERVER-73778, but for replica sets.

Before the replica set starts the internal data cleanup step, it should persist a isCleaningServerMetadata : true field to the FCV document, and then remove this when transitioning to downgraded.

We should also check at the beginning of FCV upgrade for this field and fail the upgrade if it exists.

For testing, we should make sure to test that if we fail with either a retryable error or ManualInterventionRequired during internal metadata cleanup, we fail to transition to upgrading.

When entering the CleaningServerMetadata phase we should add a log similar to the ones added in SERVER-67443 and file an Atlas log ingestion request for it