After three weeks i already finished my investigation, which leads to close my utility from public. Because i no longer thing that matching Windows sensitivity on whole screen in game is a good way. My utility was doing exactly what promised, finds the smallest deviation from Win/2D to 3D on whole screen. But I found that much better is just focusing on square in centre of image where the sensitivity is almost identical to Windows. Once we cross that orange square sensitivity will change dramaticaly from Windows. But it's OK because we need to be precise near to crosshair.
Matching Win/2D sensitivity on whole screen in game (100%) for both axis will result in 0% deviation on edges of monitor. But decrease precise near to crosshair/orange square around 0-21%.
Green line represent mouse move and position on monitor in Windows. Red line represent position on monitor in game (in easy explanation)
Matching Win/2D sensitivity on orange square (Monitor distance vertical 100%): Result in around 23% deviation on edge of monitor. But increase precise near to crosshair/orange square with max. deviation of around 10%.
So now i thing that best way for Win/2D is this:
These thoughts are related just to this game. I'm leaving here spreadsheet file with my investigation. Please note that graphs in that file do not cover all situation.
https://www.sendspace.com/file/c6fnpe