A reorg more than 6 blocks deep would probably cause more economic harm than a block that takes an hour to validate.
But there’s also the edge of a retarget period to consider. An attacker could set the timestamp of the last block as early as allowed so the difficulty increase is maximal. And then also mine the first block in the new retarget period, and reveal both at the same time. Without manual intervention an honest miner would use wall time and thus their two blocks would have less cumulative difficulty.
If you know the attack long in advance you can do that. But you might only discover the attack when it happens. That’s what I referred to with “unknown unknowns”.
So that wouldn’t discourage slow-to-validate blocks, but simply make them less harmful.