Right, but #1 is quite rare, and #2 to a use case that people wouldn't use witness data for anyways, even if there is a point to be made about the use of OP_RETURN by far being preferrable to the use of witness data, it seems a red herring to suggest that relaxing OP_RETURN standardness policy will reduce harm by increasing the number of people who use OP_RETURN over inscriptions, when there is actually strong disincentives for doing that, and the only hope is for the altruism of users.
Discussion
Right, I also don't think existing protocols have much incentive to move either. #1 isn't rare at all though, it is one of the main causes of utxo set bloat through fake p2ms outputs, which would be mitigated by using OP_RETURN.