Thank you for answering one-off even if blocked elsewhere (I block very liberally on Twitter, here it's neither possible nor needed).
The proposal you list is the second "type" I mention in my brief list. The higher order effect you mention (increasing already present offband auctions of blockspace) would be common between that and the third "type" (Much's proposal to filter out bare multisig, apparently popular among Core devs). But since Core already filters high-fee-paying valid txs, it seems to me that, at least in principle, the first "type" is the only proposal that would avoid that higher order effect completely...not the status quo. Even if I find it likely that the umprecedented move to a total lack of mempool filtering could produce way more/bigger higher order effects.
