3 years is throwing a random time at it, don't you think? Unless you have a full breakdown of what each individual should be doing and when, it's going round the house trying to guess. For the record, I've always believed in smaller teams do things in less time on the programming side, obviously reversed on the art side. However we are talking about touching up a game, not building one. GT had unlimited budget, six years and a huge art modeller team whose directive was to focus on the cars above all else.
But this isn't really about that. As for the physics engine, what makes you think it can't be imrpoved? As Weaste has pointed out the fundamentals and equations are there for all to see, it's how they are used that's important. Forza 2 had a tyre engine that it'd be difficult to improve yes, but in 3 they sorted the centre of gravity issue and going forward I think gthere's still room to manouvre on the weight balancing side of things (in 3 you just didn't seem to get the weight transfer and snap that you do when you reach the limit in real life). Besides as always, it's not just the engine itself it's what they do with it to make the game playable. To me, once you set a car up properly, forza 2 felt the closest of the console racers despite being the most faked (due to the cars not rolling over). It's about how you make the game feel to strike the right balance, and in that regards you aren't going to please everyone and unfortunately most won't beable to feel improvements if they are made.
I doubt people like me and you are going to like 4 that much to be honest. Then again, I wasn't that impressed with GT5 either.