CleanEveryN hook is typically used in various suites to clean up the data each N test runs, to avoid garbage pilling up in the system.

Once we introduced CleanEveryN hook as part of change_streams_pre_images_replica_sets_kill_secondary_jscore_passthrough we got consistent hangs in CleanEveryN runs, each failing with the timeout.

Based on the hang analyzer stacktrace, it appears that we are hanging on 
ReplicationCoordinatorExternalStateImpl::_stopDataReplication()
Job logs (ref)

Hang analyzer stacktrace (ref)