Yes, you can reorg out the blocks and actually remove them, but I

understood that you were _not_ proposing that quite specifically. But
instead proposed without reorging taking txouts that were previously
assigned to one party and simply assigning them to others.

Well, my original thought was just to delete the coinbases. But then some people don't like the idea of destroying money (equivalently, reducing the system's resolution) so I proposed reallocating it instead. I'm not sure which is better though. Deletion is closer to what the existing system allows, for sure.

Would you feel differently if the consequence was UTXO deletion rather than reallocation? I think the difference makes no impact to the goal of discouraging double spending.

... proposing the mechanism be used to claw back mining income from a
hardware vendor accused of violating its agreements on the amount of
self mining / mining on customers hardware.

I think this would not be doable in practice, unless there was a way to identify that a block was mined with pre-sold equipment. Peter points out that the pool in question is marking their blocks by reusing addresses - ditto for the double spending against dice sites - but that's a trivial thing for them to fix. Then it'd be difficult (impossible?) for miners to identify KnC blocks even if there was a strong majority consensus to delete their coinbases.

The reason I think this particular change is doable is that it should be possible to quite reliably identify blocks that are Finney attacking for profit. That doesn't generalise to any policy though. Blocks are intended to be structurally identical to each other if best practices are followed and even with the dire pool situation a big chunk of mining hash power today is effectively anonymous.