Glad to see it is so popular and hoping it is not too late for giving some support!
The pins were selected based on the manual of the PinChangeInterrupt library, which is being used in place of the pulseIn() function. It is a very nice, lowlevel interrupt library and makes it possible to use the code on more different boards. Although not every port register is supported on every board, especially on the 32u4 ones (like Pro Micro, Leonardo, etc), so that is why pins are not in order. For example on a Leonardo or Micro (not the Pro Micro!) only 4 channels can be found on the side of the boards, the rest two is on the ICSP header. It would be only possible arrange the pins on 328p (Uno,Nano) or Mega boards, but those are missing the USB-HID support.
Slight play comes from the adapter, since it have to process those values, while acting also as a joystick for the computer or tablet. Usually this shaking is very low, i would say acceptable until the diff is not more than 1 or 2. With pulseIn() i managed to get a concrete-stable input, but that was working for only 4 channels at a time.
The issue on Windows is seems like calibration is necessary, but make sure you have not made some mixing or so on the transmitter. Most of the time weird errors come from forgotten mixes.
EDIT: Ok, i've just read Freerider's comment and it is a great idea to check!