Test automation -- Beyond 6.1

I'm noting some ideas for future improvements (and keep updating in the future as I think of other ideas):

* Add ability to compare the data in the placements of shards. Use case for this: run a lot of concurrent modifications and then compare the placements of shards to ensure that nothing goes wrong.
* Consider this framework as a new kind of regression suite. For instance, as a new feature tester in the release testing, use a very simple API and add your tests to this new regression test suite. This could enable running those tests repeatedly and use it in the future.
* I don't know the methodology of this yet, just noting here not to forget.
* We could very well integrate concurrency testing into this.
* We should have an easy way of killing the workers in various ways (`kill -9`, `pg_ctl stop` etc)

The issue names no files, tests, or entry points to start from. Read the existing test-automation framework before deciding whether to pursue shard-placement comparison, regression-suite integration, or worker-killing scenarios. Done would require a defined scope and methodology for one or more of these future improvements.

Written by the indexing model from the issue text.