To what extent Wikipedia's UI/UX assessment team are aware of this and can note a strong defection from any new features even where that defection represents only a small fraction of visitors I don't know.
Worse: sufficiently annoying / offputting features can simply lead to defection from a site entirely. A major problem with A/B testing is that if the cumulative effect of features has been to drive away those preferring the path not taken, then all your A/B tests are doing is validating survivorship bias.
(I'd give my experience with Reddit over the years, which has lead me to all but entirely avoid the site, or more recently a rather drastic turn in Twitter's dynamics affecting many long-term participants, as cases in point.)
Of the specific changes Wikipedia are putting into place, the maximum line length is a style change I've applied myself for about a decade now using various CSS browser extensions. I'm not much a fan of pinned headers, though that would be less an issue on desktop than my e-ink tablet. For that, I prefer the Desktop layout generally, but zoom the screen to cut off the left-hand sidebar entirely, which works fairly well until such time as I want to view an image fully zoomed.
I do find the sidebar useful on desktop. The present Mobile layout is annoying in multiple regards though it might be useful on a smaller device (I'm using a 13" e-ink tablet, which is generally not constrained for space.)