Now entering type hell!

"Shouldn't be here" when running DecoupledAdderTester

Fixed GCD to work like tutorial
Added exerciser example to Adder in unit_tests

would be nice to have some of this verbosity controlled by a parameter file.
Ability lacking in firrtl printf to control column width makes displays much uglier

Seems like I now have a quirkly but basic system that gather all the port names (even if deeply nested) in to a Map

Meaningful error when resource copy fails.
Seems like this is exposing some deeper error with sbt and external libraries/jars.  Sometimes this error message makes sometimes the entire JVM and sbt dump with SEG fault, I don't think this is a chisel problem

Fix some nits: add that line to .gitignore that keeps coming back, do parallel testing by default

Print test stdout/stderr to console,
works for my use cases

Modify TestDriver to take resource paths instead of Files
Tested in chisel3:new_unit_tester
Tested in chisel-tutorial:chisel3prep

Added a test based version of the adder