Not exactly: in step (1) a miner discards a slow block only if a competing block arrives during validation. It will never mine on top of the old tip, it just prefers the faster (equal-work) tip. Miners don’t do anything different from regular relay nodes in this proposal.
That’s all I’m proposing.
(but if all miners SPV mine on the latest block, there will never be such a naturally occurring competing block - in order to defeat the attack some miner has to take a risk, which may be too much ask)
I don’t think we can estimate the likeness of a miner proposing a one-hour validation block. But if they do, then there could be as many as six competing blocks produced while the slowest node is still validating.
I think the question isn’t whether this attack is rare. It’s whether the risk of the attacker seeing their stale block go stale is enough to deter them. If there’s high deterrence then it may be worth some complexity.
At present SPV mining appears to be the only cause of empty blocks. But perhaps when miners also need to optimize for validation speed that would be another reason to make empty blocks.