Did a bit of testing this morning, and perhaps a clue, although I'm far from understanding it all.
It appears that Windows nominally expects the first 8 channels to be on variable controls (sticks, sliders, knobs). Channels 9 and higher get seen as being buttons in Windows. You can assign switches to "low" channels, but if they don't have a center (ie, are only 2 position), calibration can be skewed. And it helps to advance through the calibration screens if you have a switch assigned to one high channel, that will be seen as a "gamepad button" by Windows.
This may be where some of the issues came from - Windows just didn't get calibrated fully/correctly. The switched channels 5,6,7 and 8 may have confused things.
RF, for whatever reason, flips the order of the first 8 channels in its calibration screen. Not really important, but can be confusing. So, if you assign the Rud stick to Ch 1, in Win it controls the X Axis. RF then shows the X Axis as being on Ch 8 in calibration. And then in the main RF setup, you remap the X Axis control to the appropriate output used in RF (ie, Rudder). Makes my head hurt.
I haven't explored, but in setting up a new controller in RF, you have choices as to various brands of radios to start with (Futaba, Hitec, Spektrum, etc.). I suspect the main difference is the default order of the channel assignments that varies from brand to brand (AETR, TAER, etc.). What impact that has on the order things are displayed in the RF calibration and assignment screens I haven't looked at.
Sometimes just using the Interlink is the least painful option....