In tests, background tasks start executing before all sled agents register with Nexus, which can cause test flakes #7417
Description
After adding a lot of extra output to the tuf_artifact_replication background task and its omdb output handler in #7129, I ran into a problem where the output was different based on the number of sleds that had registered with Nexus before the background task starts.
Based on logs from a run, I see the contacting server nexus log message from the first simulated sled agent at 21:48:42.793Z, tuf_artifact_replication background task activation at 21:48:42.800Z, followed by the second sled agent contacting nexus at 21:48:43.179Z. The first background task fires at 21:48:42.784Z.
Getting rid of extra_sled_agents = 1 in the test_omdb_success_cases test might lessen the risk of flakes here but it won't completely eliminate it.
For now I'm planning to work around this by writing a redactor and referencing this issue in the comment.