> Hidden variables are outlawed by Bell's theorem
Local hidden variables. I see nothing wrong with non-locality; it's basically been confirmed beyond a shadow of a doubt by now. That "closeness" occurs within Hilbert space rather than its projection into position space is ultimately no problem for a PRNG.
If there was a PRNG deciding the outcome of measurements, it would have to have the mathematical moving parts exist somewhere as hidden variables. Hidden variables are outlawed by Bell's theorem and a couple of other things. Therefore, the randomness must be stateless and truly random.