OK, yes, I see the limitations of the “on face” mode and it may not be perfect for this use case. It was really meant to be used if you need a little bit more performance… In the standard “standalone edges” mode, the edges are really 2D lines, not 3D objects, in the sense that they are drawn just like you would draw a line on a screen, but between two screen coordinates that happen to be at the right position, giving you the 3D feeling. The resolution is typically lower than on a screen, and so all the lines are thicker. That’s the reason for why a bunch of lines on a far-away object ends up with too many pixels black… Switching all non-top-level edges to the “on face” mode might not be the solution either—it’s common to have models where all the geometry is in some subgroup. But maybe we can think about some other way…
A PC-only solution to try could be to draw lines smaller than one pixel wide—this should work because the PC graphic card is actually drawing four subpixels for each pixels, and averaging them out (using some clever optimization that makes it much less than 4 times slower). This avoids a lot of the “aliasing” effect that occurs otherwise and is the main reason for why the image looks a bit more jagged on the standalone Quest. So we could try to draw the lines inside the subpixels manually, half-a-pixel wide, and thus get the lines theoretically thinner than one pixel. I need to check the result and the performance.
but even there if I group together all the lines drawn on top of a shape rather than letting them subdivide that shape, the lines will shop up thick. So while that may do the trick visually, it would remove my ability to them extract individual area measurements of the areas that these lines would have otherwise created… and I do that a lot on my floor plans for instance. I hope this makes sense. Anyway, whatever you can do, it can only make things better 
Many thanks again