Any good way to unit test the command line interface?

[amoodie]

There's not really any good way to unit test the command line interface.

We test the heck out the Python preprocessor, which covers a lot of the same code that the CLI preprocessor uses, but doesn't capture the very initial parts where command line arguments are parsed, given default values, and sometimes removed from the config dictionary.

We currently don't have any real unit tests covering this chunk of code, which leads to bugs like #154