The weasel word in all this is “overweight (but not obese)”.

I think that's the whole point of the article. Lots of doctors seem to assume that all-cause mortality is correlated with BMI in a straight line, but this article argues that it's actually U-shaped with the minimum in the "overweight" range. It's arguing that these specific people in that overweight but not obese category are getting bad medical advice and treatment because of assumptions derived from observations of the group of people who are overweight or obese.