What Wheelcheck measures is X 0-100 momentarily force produces Y amount of movement. At first thought this is going to give you a linear graph so any non linearity like this makes you think something is wrong. But this isn't how we perceive force feedback, the force results in a push against our hands that we mostly resist and the wheel does not move, so already we can see there is a mismatch between the measurement and what we would want to measure as a user.
So why do we see non linear response from the "cheaper" wheels but not direct drive wheels in this test? The reason is actually that the motors are geared up and so they have a good amount of dampening due to the way the motors are geared and the slack in the systems they use. The T300 tends to have a curve the opposite direction where it has a maximum speed of movement which slows it on the higher forces as the belt stretches a little, the Logitech on the other hand has to overcome the gear rubbing and slack and increasingly high force does a better job of this.
What you would actually want to measure is the force produced with the wheel not moving not the movement of the wheel without nothing holding it. You can do that with a broom stick attached well to the wheel resting on a scale, put the force to 0,10,20...100 and read off the scale and that will give you an accurate LUT much closer to how you use it. Its not quite how you experience it because you will be moved by the force and then compensate so there is a bit of lag but its a lot closer than the free wheel spin of the wheel. You'd have to make the LUT file yourself but if you look in the file produced with a text editor they are very simple.
In practice you'll find the wheel is actually quite linear and about the only thing you really need is a minimum force, 10% or so on the Logitech and 3-5% or the Thrustmaster's. There were wheels before the modern era that were not linear but pretty anything new bought in recent years this is not the case. It may be you prefer a non linear response, personally I found my T300 was a bit too weak at its bottom end so I like to boost up the forces and reduce the dynamic range but it was fairly linear it doesn't need a LUT at all.
Wheelcheck equating movement distance to force linearity is a flawed approach, its been proven flawed for years.