But you would come to exactly the same conclusion if you left dithering enabled for all your testing. I assume 10 bit 4:2:2 with dithering is still smoother than 4:4:4 or RGB with dithering, if it is smoother without dithering.
Aren't we testing bit depth? 8 v.s. 10?
I am arguing against testing without dithering, it never tells you anything useful about your video path.
Edit: In my testing, on a C9 in PC mode, 10 bit without dithering is better than 8 bit without dithering (for banding) but 8 bit with dithering is better than 10 bit with dithering.