It was Citrea, and AFAIK they plan to use utxo stuffing only if a bitvm contract is challenged. This is designed never to occur, as it requires someone to manually create and submit a "cheating" bitvm transaction, whereas "honest" bitvm software is designed to try very hard not to let you do that. Besides the disincentive of manual work, it also carries a huge monetary disincentive: your counterparty in the bitvm contract is meant to be running software that automatically detects cheat txs and responds by taking all of the money the cheater put in the bitvm contract. (It works a lot like a lightning penalty tx, but more complicated because bitvm is complicated.) Despite being designed never to occur, this potential for extremely rare utxo stuffing by one party has been leveraged to relax the op_return limit for everyone. Which I think is silly.