Your points involve mismatches (1,0 or 0,1 -- ie., paired or coincidental detection attributes) at relative angles (ie., Theta).
The problem in constructing an LR model of entanglement is that it has to encode some sort of locality condition. This is done by assuming that events (polarizer settings and individual data sequences) at A and B are independent of each other. This is manifested in your points by calculating the expected mismatches at some Theta as being no more than twice the mismatches at 1/2 Theta. It starts with point 5., where you separate the probability at Theta = 60 degrees into the probabilities at the Theta = 30 degree offsets. This is the your locality, or, more precisely, independence assumption
Does this mean that nature is nonlocal? I don't think so. It's just that the results of Bell tests can't be understood in terms of independent events at A and B. The measurement and (assumed) underlying parameters are irreducible.
But how can one begin to understand the experimental results in a local deterministic way? Simply put, the polarizers in the joint context are measuring an underlying parameter (unlike the underlying parameter that determines individual detection and varies randomly from pair to pair) that isn't varying from pair to pair. They're measuring a relationship between photons of a pair. So, an independence assumption doesn't fit the experimental situation (even though it's a necessary constraint on standard LR models of entanglement)
. But the assumption that the relationship between photons of a pair is produced locally does fit the experimental situation (eg., see the emission model associated with Aspect et al. 1982). And then of course there's the experimentally documented behavior of light in polariscopic setups.
It all, reasonably I think, points to local determinism, as far as I can tell. So, no need for superdeterminism.