> the population robustness is people doing things differently and not dying.

No, you created a completely contrived example in an absurdly complex and dynamic system and assumed your outcome is the likely one.

You are fantasizing a completely stereotypical behavior (neurotypical people act like sheep) to say "they would all drink poisoned water and die", when the story could just as well be that after a first wave of people dying because a poisoned well, the second group of people would notice the danger and go after the others who are drinking something else.

The advantage of diversity in the population would be determined if individuals in an heterogeneous group by the same threat or pathogen survived at different rates.

> I'm saying that isn't necessarily so.

Not necessarily so doesn't stop it from likely being so, and I would stand by the idea that neurodivergent people are likely to have far fewer children than neurotypical ones.