Implement or remove multi-threading capability #227
Description
In regression_test.py, there is a threadpool worker-based implementation for launching the tests. The number of workers is hard-coded to 1. We should either implement this, or remove that, as retrieving results from futures requires more abstract code. We would need to figure out how to handle output from multi-threaded running. Also, builds are commonly multi-threaded with things like make -j, so multiple harness processes on the same node may be self-defeating.