nir/opt_algebraic: a & ~a -> 0
Also re-ordered some optimizations for better readability.
Not sure why this one was missing, but here we go :)
Only found 1 shader affected, so not really a common pattern. Or it is already covered by something else?!