After a short hiatus I just checked back in on this thread again and I'm pleased to see that the discussion is back on track.
Freebird, thanks for posting that link. It may have been posted several times already but I actually hadn't seen it before myself and I found it most informative. In particular it does a very good job of explaining exactly what causes the edge distortion on extended desktops. I had always assumed it was something to do with spherical/planer mapping but didn't realise it was literally a rectilinear translation (ie. mapping a projection onto a dead straight wall). In fact I thought people were able to increase their FoV beyond 180°, which is clearly not possible if the projection is rectilinear.
When I first started looking into this I thought cylindrical projection was an obvious solution too. In fact I'd naively thought cylindrical projection was the 'normal' way of doing things as it would seem so much simpler to just have each pixel represent a certain amount of arc, but clearly this is not what's done. I'm aware this would induce a minimal amount of distortion on monitors that are indeed flat but over the breadth of your typical monitor I would have thought this would be negligible. I am actually now wondering if old 3D perspective games like Doom may have in fact used a cylindrical projection by default. I know they use to kind of 'fake' 3D through manipulation of flat images and who knows, they could well have used a "given arc per pixel" approach to construct these scenes. It would seem the simplest way to do things.
One problem with cylindrical projections (of which 'Mercator's Projection' is one) is that while the image looks great across the centre of the canvas it becomes more and more distorted toward the top and bottom. Most of our surround displays have an extremely high aspect ratios (say 5760x1080) and a display that wraps around 120° may only extend vertically ±15° and would no doubt look fine. In fact if you were to mount your monitors in a pseudo-cylindrical shape the distortion would actually be perfect (just as a planer projection model is perfect for an extensive array of flat panels), however the great majority of players who only use a single monitor don't actually want a literal view of exactly what would be visible through a window of the size and proportion of their monitor as it would be too restrictive. Can you imagine trying to fly a real glider with your view port limited to a frame that was literally the size of your monitor as it is now? You'd be flicking your head all over the place just trying to keep a lookout. For this reason the rendered FoV on a monitor is basically always larger than the actual angular dimensions of the monitor itself. That being said though I'm not sure whether the cylindrical distortion at the top and bottom of a 'zoomed-out' display would be any worse than the current planer-projection (which I believe cartographers would refer to as a "Gnomic Projection").
Anyway, when the industry genuinely went "3D" (with real 3D ordinate based engines) they all decided to adopt the Planer-Projection model (including Condor v1.0 by the look of the ultra wide screenshots available). If the Condor 2.0 developers have decided to continue with their own engine rather than implementing an 'off the shelf' solution, who knows, they may even implement a cylindrical projection system. That would indeed be a novel solution