[tempest][swift][ceph][radosgw] debugging intermittent failures of test_upload_too_many_objects

In our clouds, we're using ceph rgw to provide the s3 and swift object store interfaces. Recently, we began running non-admin tempest smoke tests against the live clouds, and observe intermittent failures for the object store test_upload_too_many_objects test, where the rgw seemingly fails to enforce the number of objects quota. Tempest is running on the most recent master commit, cloud deployed via Kolla ansible 2023.1, and ceph currently on Pacific, deployed via cephadm. It's not a critical issue in our case, but I'd like some advice on getting to the bottom of it. Thanks! Mike Sherman