A simple whole-run testing methodology

Add test.sh for comparing against historical recorded runs.

I have not done any extra analysis on cases recorded here.  Rather,
I just want to know when/if any of this output changes.

I couldn't figure out a good way to include these tests within the
"go test" framework.  The tricky bit was figuring out how to set
unmentioned flags back to their default values between runs.  I
deemed adding an 'option' object of some kind built from command
line flags but also buildable by tests to not be worth the complexity
at this time.