Edit & Update to this ...
I've been giving this much thought. I didn't put my thoughts into words before in the post above, as I couldn't quite come up with the right words. I still haven't, not fully. But I'll try and see if I can do a sufficient job...
I would also not be surprised if there is not a very quantifiable / measurable difference between Katana SQ and THD versions that should account for why many feel it has a more pleasing presentation. Sure, there is already the aforementioned higher noise floor & THD of the SQ version. But I cant believe that is very noticable at the levels the noise H THD is at. Or that is behind what is deemed "more desirable".
I would attribute "more desirable" to be very much in the analog domain and be such things as better transparancy, better textured impact & time dependent tonal decay: i.e more 'timber-ful' music, not just of the primary tones but also of their decay in time... and how it all blends in 1/60th to 1/80th of a second beats. I daresay little to none of that will be capture in tradition measurements we are accustomed to here to illustrate a difference between the technicals of audio device design vs. the far less understood 'black art' of making a good sounding device design ... which Johan does seems to have mastered; after all, a SQ version was successfully made for a reason, right?.
(perhaps a waterfall plot [?] can hint at the time dependent behavior? but it too not capture and quantify fully just where "the magic" lies.)
Anyway, in short, I would love to see whether analytical analysis can divulge a reason for why a large segment [is it a majority?] of owners or listeners prefer the Katana SQ vs. THD ... but I wouldn't be surprised either that Amir and every reader be left scratching their heads at the data/graphs and all collectively say "that difference in the measurements can't possibly explain why there is a clear preference of the SQ version". ...prove me wrong!