Tests should be run in a pristine git repo where we control how the commits are created. This allows us to modify the contents of the git repo to perform more complete set ups to support exercising more specific senarios in a predicatable fashion.