Actually this is less of a "floating" or lift problem than it is an attempt on KE's part to correct something else.

Try moving the C.G. forward and you'll find that the plane will have less of a tendancy to float.

Sorry for venting, but I'm been seeing a lot of posts from others who have taken both my and other peoples words out of context and started posting about "poor physics" when this is not true.

There are some minor variances (which most people couldn't even spot) and some changes that affect all models, which should be corrected...

e.g. 3.5 does seem to have a wing lift value that is a tad too high. Coupled with the c.g. effect movement and you get what you are describing.

Lower the overall wing lift in the editor to 85-92% and move the C.G. forward.

This will correct what you are seeing.