BMI may be a crap metric for individuals, but not for large population samples. I guarantee you that nearly half of American men aren't at obese-level BMIs because bodybuilders are somehow making a dent in the numbers.
The author simply fails to understand that BMI is a helpful indicator for population metrics, and contorts into crazy positions to fill out a “top 10” list
But all of this nonsense nitpicking about bone density vs muscle vs fat is irrelevant when using it to look at large population cohorts over time. Looking at weight vs height is a good, easily calculable metric without needing to put everyone in a float tank.
The US BMI has skyrocketed over the past 50 years solely because we've gotten a ton fatter. There have also been plenty of corroborating studies that we've also gotten a ton weaker, e.g. studies that look at average grip strength.
There are links to the source study in that article.
Additionally, BMI is imprecise, but imprecise doesn't mean useless. If you took a random sample of people with a BMI of 40+ from across the whole population, almost none of them would be near optimal healthy weight; a random sample of 30+, still relatively few. Those few who are healthy at that BMI know it. BMI's maybe not so useful for someone who's hovering around 19 or 28, but at the extreme edges it's very indicative of body fat.