Ha, I’ll be darned! Thanks for catching this.

I’d be tempted to say that this is too small a thing to test, except … I screwed it up, so obviously it isn’t! I worry a little about writing timing-dependent tests like this: Travis is so slow and timings are so inconsistent, timing-dependent things like this can make CI sporadically fail. (Siesta’s already in that boat with waiting for Nocilla cleanup, and it's not great.) But I think your testing approach is good, so I’ll merge it and if it breaks CI we can add slop to the tests later.

Thanks!