I think it would depend on how the “weakest groups” are decided. If the weakest 10% of 2021 all died, then the weakest 10% of 2022 will be people who were stronger than 2021’s weakest 10%. All the groups would propagate up to be stronger than in previous years. Now i don’t know how these groups are drawn, percentiles is just what makes intuitive sense to me.

The linked quotes don’t seem to support your argument, unless I misunderstand? If the weakest people die, then the remaining people are expected to be more resilient to heart disease and cancer.

I think decreases in drug overdose and suicide are probably the most isolated from this effect, so I have higher confidence that those decreases are “real”. But I can imagine ways that even they might interact.