Last year I started with researching aero with the idea that there is still a lot to be gained. Most bodies right now are either sculpted by hand or made in CAD, then a mold is made and after some initial testing 99% of the time the original model will be the one being sold.
To start gathering RC data I had all the populair 1/10 nitro bodies 3D scanned and started running CFD on them to see how they would compare. After sometime I had some numbers which matched the feel the bodies gave on track.
Protoform SRS-N before converting to a solid model:
A Gurney flap doesn't work on an RC car like it would work on a 1:1 car.
In RC we use bits of plastic which channel the air upwards causing the car to be pushed down a bit, the whole principle of a proper wing doesn't apply.
First I looked into building a windtunnel but as has been said above it needs to be pretty big in order to produce some reliable numbers.
A very basic tunnel doesn't have to cost a lot, but one which is reliable and can be used to really improve downforce will get expensive quite quickly.
Personally I don't have the space to build one so I will try to do as much as possible in CFD and then start having prototypes made.
This video by Sauber F1 gives a very nice insight into the construction of a windtunnel and basic aero stuff, also how it works with scaled models: