Now that we're generally getting the HID path for DirectInput devices anyway, we can trivially look at that instead of searching all raw HID devices for one that matches the vendor/product IDs and using that to look up the HID path.  Unless I'm missing something, this greatly simplifies this code.

Only real functional change (besides potentially a lot less calling of GetRawInputDeviceList) is we're now momentarily creating the DInput interface (to query the HID path from) for some joysticks that we previously would not  have been.

Admittedly, the original SDL_IsXInputDevice early-outs on the hard-coded list of known IDs, so this is probably going to do slightly more work, but never more than when XInput is disabled anyway.  I could refactor this to keep the early-out if really needed.

Can view the diff here: https://github.com/Jimbly/SDL-mirror/commit/a8f450f66b3a0bc5835524025167cdf6cd7f738c

Patch file: https://github.com/Jimbly/SDL-mirror/commit/a8f450f66b3a0bc5835524025167cdf6cd7f738c.patch