I use joyToKey which would enable me to set my Joystick inputs as mouse movements or buttons I believe.
But if its ok with you Ill let you test, as I am sure I would mess it up somehow
Did some more testing and it's not as promising as I initially suspected.
Much of the improvement to fine control was simply a reduction in effective sensitivity that can be duplicated with lower latency and more precision with an actual joystick curve than mouse emulation + relative mouse. Once I adjust everything to allow for full magnitude inputs, it was more or less just a slightly laggy version of the standard stick controls. Essentially, the mouse emulation was just another filter layer and relative mouse was just turning the emulated mouse back into a stick.
To get an equivalent enhancement to stick utility there would need to be some sort of temporal, rather than positional, dynamic sensitivity, which I imagine would be very difficult to implement without destroying the advantages inherent to the stick (being able to hold an input without degradation, yet not having to worry about losing one's zero point as with a mouse).
I'm mostly back to my original position of the mouse and stick being different, but broadly equivalent controls, even if relative mouse is kept. And I still think the core issue is the silly and arbitrary combination of FA On/Off capabilities that try to balance each flight mode against each other rather than do anything rational, kinda like the 'planes in space' flight model in general.
However, it is curious that the mouse actually has an in-game power curve, but other analog controls only have deadzones. This gives people using third party tools for curves a major advantage over directly connected devices with linear response.