I took some of the courses you mentioned at BYU BTW, but it's been more than 20 years since I was a math major trying to decide if I should go the PhD route. I also probably implement some of testing you're discussing in the clinic these days if it's what I imagine you're referring to.
But back to the point. I'd take a self trained statistician any day. And they can accomplish a lot of work @$60k per year that is actually directed at producing a clinical objective. There is a lot of wasted effort in industry but they do a good job measuring it and eliminating it. Can't say the same for academia.
I agree with your assessment of the worth of undergrad classes though. Nearly a complete waste. Employers should move toward a certification process rather than degrees.