The way I've heard it, BMI can be inaccurate on the individual level, but it was created to analyze populations more broadly, where it can be a somewhat useful heuristic. I would guess athletes with obese BMIs are a small enough portion of the population to not make too much of a change to the population-wide statistics.
Beat to the reply by a few other people on the accuracy across populations. The only thing I'll add is that BMI is a much more accessible measure. The more accuracy we want, the more invasive and expensive it is to collect. Just about anyone who's ever been seen by a healthcare provider has their height and weight taken so it's a very easy metric to capture across an entire population.
Most likely done based on BMI because most methods of measuring body fat are not easy to access and have significant variability. I agree with you that BMI is not the best method, but it's probably the best default because it's easy to measure.
If you are ripped af and most of your body mass is muscles then yes your high BMI can be safely ignored. For everyone else, it's definitely a "good enough" reflection of their bad habits.
BMI can be inaccurate for an individual. That's because it was never designed as a metric to be applied to individuals, but rather to populations. This page is an example of BMI being used appropriately.
On a population level, BMI actually underestimates obesity https://abcnews.go.com/Health/Wellness/bmi-underestimates-pr... so the real obesity percentage is probably even worse than 42.4%