This diagram represents how rectilinear projection works - it shows a progression of FOVs. As the FOV decreases, the circles get bigger, the circumferences increase and the arc between the bounds of the monitor edges becomes flatter. Then as you approach 0 FOV, the circumference approaches infinity and the arc, and even the rest of the circle, becomes completely flat, until it is considered 2D. 0 FOV is both 2D and 3D.
And even though we can't define 2D in terms of cm/360, because the circumference of 0 FOV is infinitely long, its "sensitivity" can be defined another way... the chord length. If the FOVs all share the same chord length, then the length of the chord also determines the circumference for all the fields of view. It's in the diagram, so the proof is in the pudding. So whatever your 2D edge-to-edge sensitivity is, you can use trigonometry to convert it to a 3D sensitivity and vice-versa.
And from testing, this method yields the same results as 100% MM and the gear ratio method, but with the addition of not getting an error at 0 FOV. It's just another way to look at things, and only further solidifies the fact that 100% MM would be the only true method, if it weren't for distortion.