3. They died more and if anything was gonna make them sick they wouldn't have survived till old age so the survivorship bias is stronger.

It's sort of this, but also the fact that risk factors are probabilistic and there are, to put it bluntly, a very large number of people. Some might theoretically have a small mutation that makes their system more robust in some indefinable way, but mostly it's just a matter of there being so many people that some of them are going to win all the "risk of imminent death" dice rolls for a very long time.