I only use 1x1 binning for RGB. I did try some 2x2 a few times, but I found that the data just was not as good as 1x1 for my purpose. I just am not a fan of 2x2 and then adding it to 1x1 luminace data.
Yes, it was for testing purpose to compare "real" Luminance with "Synthetic" Luminance. To an average person, it was difficult to tell the difference but when I was comparing the two by blinking in between the two, "real" Luminance was the winner.
It's possible if the original RGB subs were capture in binned 1x1, the result would have been different but you would have to capture significantly more RGB subs to nearly match with "real" Luminance. That possibly explains why Synthetic Luminance was not as sharp due to RGB being captured in binned 2x2.