Add "--stop-on-failure" to reference and test commands

Currently debugging invalid ready code is pretty tricky. For example, it's possible for `waitForSelector` to timeout, but there's no easy way to determine what test case actually caused the error. We end up hacking our test case files, removing tests until we find the one that is broken. Of course, this is made more difficult if the failure themselves are spurious.

PHPUnit and Behat both have a `--stop-on-failure` option, which does exactly what it says. It would be great if backstop had an equivalent flag, and when failing show exactly what scenario failed and the line it failed on.