Run Python test suite with hook_tiered.py using s3_store instead of dir_store, and optionally add to Evergreen. Optionally, because test cleanup (bucket removal) could be tricky, and if we don't have it right, we don't want every evergreen run leaving objects/buckets around after successful runs. We might have to create new bucket names or prefixes for each run (so simultaneous runs don't step on each other). How long do we keep objects/buckets for after successful runs? After failed runs? A lot of unknowns, and many issues can be spun off into future work. Some of this may have been examined during the s3_store intern project.