I have, as of yet, not seen a single attempt to support your "non-ideal body types are more likely to sustain injury" with actual real-world data. I haven't even seen an attempt to strictly define non-ideal body types, which allows for a lot of goalpost moving. As a result, I treat your theory with extreme skepticism.
You're not going to find an objective measure. Durability is a combination of luck, toughness, running style, usage, and body type. BMI is a pretty good starting point for determining body type, but it's flawed because it doesn't provide any information about the distribution of a player's weight, which is important. All else being equal, a top-heavy back is going to be less durable than a back who carries his weight in his lower body. The best support for the "ideal body type" argument is the fact most of the RBs who have had great careers in recent history have fit the mold pretty well. Guys like Ricky Williams, LaDainain Tomlinson, Marshall Faulk, Barry Sanders, Clinton Portis, Shaun Alexander, Jamal Lewis, Frank Gore, and Edgerrin James are all built more or less the same. It's probably not a coincidence that Trent Richardson and Doug Martin are physically cut from this same cloth as well. Those guys are actually the exact same height (5'9.1"), and their weight is only separated by 5 pounds. The NFL clearly has a preference for a certain type of RB. It drafts more of them in the first round. More of them go on to great careers.To be fair, we've also seen some thinner backs like Darren McFadden and CJ Spiller earn very high draft slots in recent years, but neither has yet demonstrated the ability to accumulate a Tomlinson/Edge/Ricky workload over multiple seasons. The same goes for Reggie Bush, who has a career high of 216 carries after 8 years in the league. Jamaal Charles is hovering around the same territory right now. Chris Johnson is really the only guy with a non-traditional build who has shown the ability to handle a high volume of carries over multiple seasons, and he has a few subjective factors working in his favor (running style, distribution of weight, super freak athlete). I don't think BMI is a perfect indicator of body type, but I'd be interested in seeing a study that tracks all backs who were drafted within the past ten years and determines the average workload as it relates to body type, factoring in draft position (because a non-ideal first round RB is probably going to have a higher expected amount of touches than an ideal seventh rounder due to differences in talent). I'd venture to guess that a study like this might corroborate some of what I'm saying.