The CHSH proof of S<=2 for hidden variables supposes that integral over all angles "phi" of probability of passing/interacting of spin/polarization/etc having angle "phi" with detection axis is 1 (i.e. either no loss or the loss is proportionally distributed over "phi"). For example like the integral of Malus law's cos^2("phi") produces 1. That is in the theory. On practice i haven't been able to find experimental confirmation of Malus law for single photons. Practical loss of high "phi" photons on polarizer - the cut-off or significant dampening of the trail of cos^2 - would easily produce S>2 for hidden variable "polarization" of photon. Moving cut-off to pi/4 (almost like Einstein suggested, the difference is that he preserved probability 1 which isn't necessarily true for practical experiments) even produces the maximum S=2.8 for hidden variable theory. The same applies to spin as well. Notice it isn't a detection loophole which is about detecting after polarizer or after spin measurement, it is about the measurement/interaction itself.

In the posted article the entanglement between the electrons exists only in theory - 2 different physical objects - the generators - receive the same random number, and the connection ends there. They could have used 2 completely separate random generators, and after completing of the experiment to use only the events where the 2 random numbers were equal. Absolutely no physical connection, yet I bet they would still see the "entanglement", entanglement post fact um, propagating from present into the past.

At least in downconverted photon pair case one may try to believe in some entanglement magic because the photon pair is "made" from one photon - ie. there is real physical connection exist at some moment in the photons' past.

In short - to state Bell violation here one have to experimentally show the single electron spin Malus law equivalent.