I tried lots of tests last year with ride heights after hearing about this "backwards" issue. I never did get my head around it properly, but it was clear that adjusting only ride heights caused the physics to react in an unusual way. It was also hard to ignore that some drivers really got to grips with this unusual effect and got some amazing lap times. I never managed to make it all work for me though.
The oversteer and understeer characteristics I iron out with spring rates, dampers, ARB's, camber, toe, LSD and brake balance. I find it much easier to target the handling characteristics I want by using all of the other components, and simply use ride height to provide the amount of suspension travel needed to let those characteristics happen and keep my tyres in contact with the ground. That always results in having a slightly raised rear end for me, as you might expect to see in the real world. Coincidence perhaps, but my feeling is that this is how the ride height model was intended to work. Take the model outside of this window and it provides a different effect. I dunno. Whatever works for you I suppose. I should add that I do quite like a lot of suspension travel front and rear for my driving style. More than average I think.