It is actually an A/B test inside an email program, not a smart campaign. I know the way you are suggesting, that’s how I use to setup our A/B tests. Doing it inside the email program is easier but it appears that once you go live with it, it has to run its course. Which is a bummer because we realized the test was faulty and need to be adjusted/restarted.
It would also be cool if you could do multiple weighted winning criteria. For example, the winner for a subject line test could be based on opens and clicks. Where opens with clicks are worth more than an open with no click.