Agreed.
The main issue to how to couple transaction shunning with something effective to stop profits for miners through out of band payments. This is a both or none type scenario.
Discussion
I don't think you need to go out of your way to prop up a libre relay network. If you run knots, there's no difference to you whether a spammer uses core or Mara pool. Its gonna be negligible on fee rates (and thus miner centralization) unless theres more spam as a percentage of transactions, than miners willing to mine them as a percentage of global hashpower.
Mara, in running a private mempool, is gonna have to take personal responsibility for what they mine, and their gains from mining spam will probably be outweighed by losses from filtering out non OFAC compliant TXs.
Making unusually profitable transactions available to all miners in the goal.
If the current fee rate is 10 s/b, a spammer isn't gonna pay 30 s/b to mara. They will just pay 11 and broadcast to the libre relays. The more of a gap there is, the more incentives for miners to defect and mine these span TXs. It's market incentives all the way down.
Which is why relaying spam makes sense
What I'm saying is, let the market handle it and cast your vote based on what you wish to see on the chain. Don't make up complicated hypotheticals to explicitly do something against your values for the greater good.
And what Iām saying is that doing so causes the problem that nobody wants.