I don't think you get to say "just design a universe such that X" without actually producing one. In this context, that's a circular argument, since the possibility of such a universe is precisely what we're debating. And lots of things that look reasonable turn out to have inconsistencies buried deep inside. If you're not familiar with computability, "write a program that figures out if another program will run forever" sounds possible at first. You can make specific counterexamples to my argument, but just saying "it must be possible" illuminates nothing.
Anyway, I'll try to do better. Basically, in a universe in which "evil" is physically impossible, you may have "free will" in some sense to choose which good, but you cannot choose whether to do good, because all possible choices are "good". It's specifically freedom to choose good or evil that's the key ingredient in "good" being a meaningful concept, or at least it's consistent to so assume. You're back to the pretty clockwork.
I don't know if it's possible to construct a bulletproof argument here. Shaky foundations are an occupational hazard of metaphysics, for both sides. Saying "well, he's God so he should be able to figure it out" works just as well as "he's God and couldn't do it, ergo it's inconsistent". The best we can do is show our positions are consistent under some set of assumptions.