You didn't say, "
even if it's the same for all cars (and parts / modifications added to cars), it'd be an improvement over what we already had" did you?. Mr. "Doitrightthefirsttimeordon'tdoitatall"?
What is the world coming to when we can't count on you for advocating getting things right the first time!!?
Baby steps. There is a difference between using CFD to predict off-axis lift, drag and CoP etc. vs., say, just using a cosine drop-off for yaw's effect on lift. The former is "doing it right", regardless of whether it can be done per car, the latter is just ignoring what is actually going on, even if it gives a "good impression". At least, one is "more right" than the other.
I'd prefer it if they could use (simplified versions of) the meshes they've modeled and run a few characterising simulations per car, and it's very possible they have done, but it's not necessary to move things forward given the general state of aero modeling in games. Ideally they'd test every car on track, in the windtunnel etc. as well, but that's just not practical.
I never expect anyone to get things right
first time, though, just that you use the method that gets you to the "right" answer, or close enough to it, with enough feedback over time, using the tools that are available and within practical limitations. It's effectively just an extension of science ("knowledge") - you build up a picture from the basics and progressively add detail. Then you put your engineer's hat on and start seeing which bits of that exquisitely constructed image of reality you can happily amputate and still get the results you need.
If you start with the wrong basics, though, you're better off not bothering, unless where you're going with it is inconsequential.
