Bill, you know exactly how this happens. Say in your shop you and a co-worker are told to copy a Deuce grille shell apron, based on a genuine part on a Deuce in your shop. You can't tell me you both will end up with the exact same piece when you're both finished. How exactly, without any high-tech computerized measuring to you measure and recreate the exact shape of the part? You can take measurements for 24 hours straight, and even share measurements with your co-worker, but the two finished parts will never be identical, and it's simply not realistic to think you can measure every single point on a surface as complex as an apron, or harder yet, a crowned roof on a '56 Chrysler 300.
I get your point and I understand the frustration of seeing about gross errors being missed along the way, but scale model masters (at least at Revell, and it appears Moebius, too) are made by humans, then scaled down (one of the circa 2000 issues of SA had an article with images covering this very topic), and yes, pantographs were still being used for this process. I recall Lindberg's '66 Chevelle SS and Revell's '40 Ford coupe were two of the featured models.
Let's say for discussion's sake Revell has the ability and opportunity to 3D scan an original '32 Ford Phaeton in Jay Leno's collection (no idea if he has one, just an example), then your shop is contacted because it also has a gennie '32 Phaeton. Let's say Rad Rides by Troy has a third original '32 Phaeton, which Revell will also 3D scan, so they have three original cars from which they will obtain 3D data. There is NO WAY all three will share the exact same coordinates in the X, Y and Z planes, so which measurements do they use? Car A's LR fender has a 1/4" larger wheel arch radius than Car B's, and Car C's is 1/16" smaller, Car B's cowl's max width is 3/16" wider than Car A's, and Car C's is the same width as Car A's, but has 3/32" less crown, and so on. There HAS to be a human element involved when these models are designed, and that means interpretation on the designer's/model maker's part. No single "perfect" example of a mass produced car exists in the real world, therefore, there can never be a perfect model, especially when the model is viewed and interpreted by many different people, each of whom sees things differently than the next person.
We also have to consider the effect of proportionately decreasing the measurements, such as when a 3/32" drip molding radius is shrunk down 25 times times to .00375". That type of precision isn't possible at 1/25 scale, neither by injection molding nor 3D printing, so again, there's going to be a compromise. They add enough thickness to allow the detail to appear on the 1/25 scale model, but that scale incorrect drip rail trim may make the roof look flatter to someone's eyes since the model's drip rail trim to roof proportion is no longer true to the real car's.
Again, I understand how things should be in an ideal world, and gross errors should not be tolerated, but human involvement can never be eliminated at all points from idea to finished product, so for that reason alone, perfection is not possible.
Some flaws in your otherwise excellent and quite correct message here: For starters, in the now-largely departed era of hand-carved tooling masters--before the digital revolution reached the design studio and tool shop, the camera was the researcher's principal tool. Add to that some very basic hardware store tools: Folding carpenter's rule, a pair of measuring tapes (steel tape where it can be used, and a non-stretching dressmaker's cloth tape measure for stretching across a painted surface (funny how owners of real cars don't want a steel tape on their thousands of dollars paint jobs!), a notebook and ball point pen for making notes about this-and-that. Even a common stepladder can be handy IF one has a vehicle with room to carry one! In almost innumerable ways, 3D scanning can take much of the work away from such "analog" methods, but still, anyone researching a real car for creating a miniature is wise to take hundreds upon hundreds of pictures from every angle imaginable (and with high-quality digital cameras, this has become less expensive than a cheap suit!). Why so many photographs you might ask? Well the answer is quite simple: Photo's that show the crown of a roof from all angles go a long way to confirming (or denying) the accuracy of what the digital 3D scanner might see, and can aid in making such small corrections of shape that make the miniature believable to the human eye (bear in mind that our eyes, given their "binocular" vision, with the visual field of each eye overlapping that of its neighbor can make an EXACTLY accurate shape in miniature appear not quite the same as the full-sized 1:1 scale original. As for such small details as a drip molding, that can be tooled exactly to scale, but it can also become so small as to make it simply disappear under even the thinnest coat of paint (one of my pet peeves has always been scripts and badges engraved so faintly in scale that they simply disappear, when only a slight adjustment in just the depth of those details will make them pop out in a decently done paintjob, even with a rattle can).
Now, with measurements: Never a good idea to rely on equipment to give exact dimensions crack out of the box. Wherever possible, measuring tools need to be used, to establish firmly what correct height and width should be. A carpenter's rule (and I have a couple of them!) having every other inch blacked out, and laid across say, the '32 Ford firewall you mention, will give, instantly, in a photo,the width of that panel--and it's pretty easy to figure that down to a fraction of an inch just from the picture (bear in mind, at .040" to the inch in 1/25 scale (or 1mm to the inch--doesn't much matter) a scale quarter of an inch comes down to .010", which is pretty nearly the tolerance any modeler can achieve with a needle file or 400-grit sandpaper. That same, marked up carpenter's rule, or even a dressmaker's cloth measuring tape, similarly blacked out every other inch, can clearly show the placement of chrome trim, scripts, badges, even door handles and lock cylinders; in addition to confirming the height of say, the windshield, side windows, even the lengths of such.
With engines and all the other greasy stuff, it gets more difficult: While of course, there should be little tolerance for major dimensional errors, simply finding say, a 392cid Hemi OUT of a car isn't always as easy as it may seem. It would be great if they were just sitting around, on engine stands, just waiting for a model company researcher to come around, measure and photograph it. Every once in a while, one does find an engine in such a situation: The 308 Twin H-Power engine that Moebius was able to reference IS on display at the Hostetler Hudson Museum in Shipshewanna IN, about 35 miles or so from Dave Metzner's front door (I have a full walkaround photo spread of that engine that I took on Labor Day Weekend in 2011, and it helped immeasurably when I bullt my two Hudsons (and will get the call once again, when I decide to superdetail one!).
While I cannot vouch for what happens at Revell, the Moebius tooling mockups are done IN 1/25 scale, not in say, the old-fashioned manner of 1:10 scale which would require pantographing down to 1/25 scale injection molding dies. That alone removes a huge set of steps in the tooling process. In addition, they do have the appearance of having been carved out by CAM, with probably some small detailing added by skilled hands, but I won't speculate as to what or which.
Enough of that. One thing I learned when doing major body conversions for resin casting was to ALWAYS evaluate shape and countour by studying the body shell during that work by holding it as nearly exactly the same angle as the car in a picture which I was comparing it to. Additionally, I also closed first one eye, then the other, so I could see the body shell in my hand as close to the manner in which the camera saw it--one-eyed, or monocular. Given the curvatures of any car body, that's the best and most accurate way of judging shapes of the model that I was ever able to come up with.
Economics do necessarily come into play with the development of any product, be that a model car kit or anything else you can think of in our world. There are definite limits to the money that can be spent, at all stages of product development, and tooling costs are but one factor. Time constraints also do come into play, just as they do with the real car.