Seems that Rtings rate the Edition XS very highly overall. Rightly so!
Don't take the bare numbers on distortion too seriously, they don't mean that much. If they were as low as usual then one would say, checked o/k. Now as they are not that low, but stand out in comparsion to the competition by a fair amount, and those numbers vary from sample to sample by nearly the same, one may be tempted to risk a closer look.
In this case first forget about Rtings. The presentation of their measurements is not designed to inform on details. It's fair enough in the regular case for sure, but not here.
You say 'I'm o/k with mine.', which is nice to hear. But is Your's representative for the whole batch, and if so, what are my chances to pick a sour one anyway? If Your's an outlier in that it exhibits that vast distortion as we see with the prohibively costly He1o-Rp discussed here, at least I hope it's an outlier, what can we learn from Your statement?
Is Your XS type like this, and if so, what would be the consequences for the sound profile, considering the use profile, when considering that distortion numbers are just a coarse indication for an underlying irregular behavior? Technically speaking.
Ps: wasn't it good practice in engineering science to acknowlwdge the unexpected, scrutinize the
model** an behalf of that, here 'mildly nonlinear' - which it is not anymore by a long strech, and ask for some rational explanation? To mitigate the psycho-acoustic problem by talking it down professionally may be science too, but not engineering.
**on what '
model' means in this context: model concept, an idea of what a thing is; before one puts numbers to something is has to be understood what its nature might be, its dimensions just with a box - its existance is identical with its descriptors, width, depth, height, and only once these parameters of a mental map are determined a measurrement makes sense; same with "harmonic distortion" as that is only one of many descriptors for an underlying real thing, a process basically, so "HD" is not a concept in its own-- it is a proxy if You will, an approximate only valid it the numbers are small; once they shoot up, give up with the simplified model and look somwhere else
-----
Finally I did, and found with my meticulously selected sample of another Hifiman, HD ~ 1% at 114dB, no issues with tone bursts, small/wideband intermodulation, and other nasty stuff, but dunno what 10% HD would spell, of course.