If I had to guess, I'd say the mechanism behind that is it's expensive to be antibiotic resistant, and when you revert to normal times the mutation becomes a burden. That's serendipitous, but if it turned out the mutation was OK in either case we would be screwed.

This is not really the best way to cognate about mutation at the bacterial scale. A better way of putting it is that every possible simple mutation is happening right now. If a mutation were truly neutral, it would already be fairly present in the population. It is possible that a complex mutation could emerge with these properties, but the likelihood is much less that the likelihood of many other things that would screw us more.

All that aside, these antibiotic regimes have been carefully designed by people who have already thought very hard about these problems. The single-dose model is designed to produce a significant response with a non-durable mutative result in the population. There's an awful lot of literature about the design process for third-world medication (see e.g. tuberculosis treatment) if you're curious.