This math is wrong off by a factor. Based on Daniel in SD's chart, that would be 185,000 - 370,000 deaths (0.1 - 0.2%) in that optimized scenario. And I get your point about how improbable that scenario will be to play out, but Kim's point was about human psychology, and you absolutely can NOT contain people. This lockdown method will leak like a sieve.
Those darwin award candidates in Kentucky (even by the worst estimates, 7 out of 10 will live, with best case being well above 9 out of 10) will be immunized and back to normal far faster than us lockdown folks. Yes people will die, but it seems they'd rather deal with that then to have people die from treating themselves with aquarium cleaner.
Thanks to decades of modern medicine and society, natural selection no longer selects only on ones physical environment, but psychological and economic elements as well. We might become the new white moths fluttering on the dark bark of an asp tree. Over-caution can be deadly too in certain situations.