Well,CAD files are CAD files, and they may or may not be accurate--after all,one of the very first lessons I learned in Computer Science 101 in college (Summer of 1969 if that means anything) is the acronym GIGO, which translates to "Garbage in, Garbage Out". Does that mean that all computer data as it relates to scaling down a real car to make a model of it is garbage? No, it does not. However, more often than not, such information simply has to be fed into the machine by a human or humans, who are actually translating what they (or other humans) have seen or observed. Simply put, the computer is but a tool (highly technical and perhaps complex, but nonetheless no more or less a tool than a hammer or a screwdriver).
Bear in mind, that a model car (and the illustration starting this topic involves TWO real cars, the design and styling of which were done long BEFORE computers were anything more than electronic calculating machines! Unfortunately, model companies (and those within those model companies who do the actual work of product development) don't have the multi-millions of dollars worth of equipment, along with a very large staff of stylists, engineers and craftsmen at their disposal that the likes of Ford, GM, Fiat-Chrysler Automotive (or any other automaker), but have very small development teams, anywhere from 1 to perhaps a dozen in product development, with computer skills and computers of course, but it is NOT the same thing. So, the human equation steps in, and depending on the commitment of a particular model company to scale fidelity, up or down, does affect the final outcome. In all this, computer technology, at the model kit development level, simply has replaced the draftsman at his drawing board, the pattern-making tooling mockup sculptor with carving tools in hand, and the old-fashioned toolmaker at a 3-dimensional pantograph milling machine. The best that can be done with cars manufactured before digital technology such as is available today is simply old-fashioned detective work--the upcoming '65 Comet Cyclone stems in large part from perhaps 400 photographs taken of a real car, perhaps 30% with specially marked rulers and measuring tapes laid on or next to portions of the body to give exact dimensions of those parts--but unless one is referencing a rust bucket, it's just not possible to measure each and every body panel.
As for access to 3D CAD files, even that can be problematic, depending on the depth of what a real automaker might provide (I still remember, when at Playing Mantis from 2002 to the end of 2004 (doing product development for Johnny Lightning diecast (1/64, 1/24, and 1/18 scale) we got a set of body loft drawings from Ford Motor Company for the 1964-66 Mustang which were TOTALLY useless--they were FULL SIZE drawings, the same actual size as the real car! Try working with something like that in a 12' square office sometime! And yet, when I was assigned to develop the Johnny Lightning 1/24 scale 1957 Ford Courier Sedan Delivery, Ford's then diecast model subsidiary and licensing office was able to send me a set of 4-view basic line drawings of that car, which along with the nearly 100 photographs we'd already found, were enough to make a very credible 1/24 scale model. As with any car designed before modern digital imaging, scanning, even CAD meant, and still means creating all of that data. And then, as hinted above, there is the level of commitment to making a correct scale model--some companies and some management teams (the former often outlive the latter!)--can make or break the final product. Now, as for the two differing Mustang model kits which started this conversation are concerned, they were done by two different teams, by competing model companies, with all the opportunities perhaps with wildly differing budgets. But on the flip side of it, there are numerous instances where a modeler has taken a body shell from Model Company A, played with it on the body shell from Model Company B, and found that it fit with little more than a swipe of a file here or there, or perhaps adding very thin strips of styrene to enlarge the hood the very small distance needed for a reasonably good fit.
In short, Harry, the process as nowhere nearly as simplistic as you seem to suggest!