Ya, NGons are just a modelling and storage convenience. When modelling in wire mode, your mesh looks a lot cleaner, that’s pretty much it.
Hardware acceleration, be it OpenGL, DirectX, etc. use trigonometric methods of plane-geometry. Two methods for fast mathematics: a single vertex and a normal defining the plane’s orientation, or declaring a triangle…a plane defined by 3 points, which define both the plane and the face perimeter…makes it easy.
Defining any more than 3 vertices for a face breaks the logic just described since a fourth, or nth vertice need not lie on a single plane. Sometimes you’ll notice in any 3D program that uses FGons/NGons, that there are flaws in the shading. This is because inevitably, the software has to break your NGons into triangles to calculate the shading and rendering. Collision detection also requires plane geometry.
I see it as nothing more than a way to keep your lines in wire-mode from becoming a headache to deal with.