I apologize in advance for some of my ideas based on your post, because you are obviously both knowledgeable and experienced in using this particular combo of TT and tonearm, but here goes:
(1) I am sure you know this, but there is one thing more to setting up any cartridge on any linear tracker that you did not mention, which is that the stylus tip has to sit exactly on the radius of the LP such that the arm wand is perpendicular to the radius. Else you will have wow (which could be a source of the distortion you hear). Did you take care to get that right? (Most likely you did and just did not mention it.)
(2) Bubble levels are nice but not the absolute most accurate way to check that the platter is level. Better is a linear carpenter's level, used to check level across the platter in two planes at a 90 degree angle to each other. (I am assuming you used one of those small round levels that has a circle in which to center the bubble. I use one like that only to "ballpark" level.)
(3) Is it really the best idea to have an air bearing tonearm on a spring-suspended TT? Any relative motion between the stylus and the platter is a no-no, and in my imagination I am thinking that could happen with a passive air bearing arm like the ET2 on a spring loaded TT, and maybe that is why you are not getting optimal tracking. I personally don't like the idea of increasing VTF just to improve tracking when you are already near the upper end of the range recommended by the manufacturer, but I would not say it's "wrong", especially if it works.
(4) "Distortion in both channels" probably has nothing to do with skating in this case. Also the HFNRR instructions on how to attain "correct" anti-skating (which does not exist) do depend on the acuity of the listener, probably not very accurate which probably doesn't matter. Another way to use their mono AS band would be to make one channel 180 degrees out of phase with the other (by for example swapping wires between hot and ground at one speaker.) Thus the signal in one channel would cancel the other when AS is set optimally, at least for that frequency at that segment of the LP.