Because it's not very accurate in transient cases, or when following another car (as the wakes of cars are vastly more complicated than the idealised CFD modelling). But I don't have hard proof, no. Last year a friend of mine was attempting to model a tyre in CFD, and spoke to various F1 teams about it, for his masters degree. He was told (and they might have been lying) that it's not something they can do themselves accurately, so anything he did would have to be fairly gross approximations. If F1 teams/tyre manufacturers can't model flexible tyres in use with bodywork around it from an aerodynamic standpoint, I can't see LFS doing so.
Yup, in BF1s to UF1s, and I didn't see, or be involved in, any crash like that. A few knocks, bumps and spins at T1 , but nothing I'd call 'bad netcode'. And I'm on crappy, laggy wireless at home