Another thing to note that current PoC has a possible collision for hashes because tokenId’s being offseted for (0…3) to define different kinds of requirements.
Right now I think that best way around that is to add these offsets in leftmost byte (msb) of the token id. That way exact signature hash match (msg.sig casted to uint256) can stay the same, but incrementing can be done by simple OR bitewise mask.