sharding_auth and sharding_auth_audit under TSAN caught a few data races (see comments for BFs). However, according to the analysis in SERVER-93653, we don't actually see any other failures in sharding_auth or sharding_auth_audit that aren't already seen in sharding.

We would like to delete sharding_auth and sharding_auth_audit but due to their incidental coverage (in catching a few data races) we don't have the confidence to delete these suites.

It would be great if we can add explicit testing for the coverage that is currently incidental.