Interesting, but if someone attacking ECDSA with a quantum-computer doesn't make much sense to use quantum-resistant hashing along with ECDSA (which is very likely getting broken) because the attacker can take everything regardless.
Old nodes won't understand it anyway. Which means, that only quantum resistant nodes will see it. And for that reason, ECDSA verification can be just extended: first, ECDSA signature will be checked normally, and after that, quantum data can be hashed, and checked, if it matches R-value of a given signature. And then, new limit don't have to be based on block size, but on commitment size instead, and based on existing sigops limit.
Regarding the old nodes: There isn't much nodes that are not segwit compatible. I think in a few years (like it is proposed to switch to quantum-resistant algorithm) we can completely ignore old nodes and simple "kick them off" from the network.