Once #165085 has landed adding variable shuffle mask handling, we should be able to update interp__builtin_ia32_shuffle_generic to handle 2 arg (1 operand + 1 mask) cases as well and replace interp__builtin_x86_byteshift / interp__builtin_ia32_pshuf / interp__builtin_ia32_pshufb

Same for evalPshufBuiltin / evalPshufbBuiltin and BI__builtin_ia32_*_byteshift handling replaced with an updated evalShuffleGeneric