This test uses the same collection for every test case, making it slightly confusing to diagnose test failures in the CI system. It also drops the database (which has an implied majority wait) before running each test case, which is unnecessary and inconvenient for other tests that may need to run concurrently (e.g. parallel test suite).