It's hard to know, right? I would guess the average performance for my version is worse, for the reason you mentioned. But the worst-case performance for your version is worse, because in theory that loop can spin and spin until the random number generator picks a different fruit.
But really, the performance costs we're talking about aren't important, if they only happen once per human click interval. I'd favor readability... if I could just decide which one was more readable.
... Anyway, I just think it's fun to see multiple ways to do things.
We're a place where coders share, stay up-to-date and grow their careers.
We strive for transparency and don't collect excess data.