As many a good question, this one is really several questions in one. Let me do some untangling.
The general question of editing answers for "improvement"
Is it good when people edit other people's answers for what they believe to be improvements? I think this is subtle and (to my knowledge) not really decided in general terms on m.se, but most of us have seen edits that are clearly good (e.g., fixing non-compiling LaTeX) and edits that are clearly bad (e.g., simplifications that make the proof wrong). The truth is somewhere in the middle and everyone seems to be eyeballing it. My personal criterion is: do I think I am helping bring the author's intent out better, or merely improving my personal view of the answer? Thus, e.g., it is fine to fix LaTeX and add a few words here and there to disambiguate or explain a nonstandard notation, but I would never (e.g.) generalize an answer stated for the integers to an arbitrary commutative ring. There are many shades of gray here, but at least the criterion is something to guide oneself by. If one really wants to go beyond it, one can always comment.
Incidentally, math.SE needs not concern itself with general SE principles here; we are somewhat unique in that our answers don't rot like code nor change like the weather. Besides, I believe we mathematicians take more literary pride in our writings than the typical programmer (at least as concerns the kind of code snippets that make up most SE answers; programmers pride themselves on bigger projects). A widespread attitude of viewing answers as interchangeable and depersonalized will put some of the best answerers off.
Exoticism and the "improvement" at hand
Independently, as to the specific "improvement" at hand, very little justification has been given to convince us that it is an improvement indeed. Jeremy Brazas, in his twitter thread, limited himself to a rather circuitous guilt-by-association argument:
I do NOT believe the name "Hawaiian" in "Hawaiian earring" honors Hawaii or native Hawaiians. The whole "you should be honored that you're even involved" response to Hawaiians is a commonly used tool, often by well-intentioned people, to perpetuate stereotypes and maintain the status quo for marginalized groups. Continuing to associate "otherness" or "exoticness" with a marginalized people for no reason other than tradition is not ok.
So, yes, the "you should be honored" supposition is unfounded, but why should anyone be offended either? What is the causal chain starting from a use of an ethnic name on a mathematical object to any objective harm inflicted on anyone from the respective background? And in the absence of a theoretical justification (preferably without 1 year of philosophy background required), is there any experimental data that is not built entirely on self-reports by a rather narrowly selected group of academics in the humanities?
Similar arguments could be made for "Russian" roulette, "Chinese" whispers and many other artefacts of our culture. Russian roulette, at least, has successfully made its way "back" into Russia and has become somewhat of a point of pride among some Russians, who appreciate a catchy metaphor for what they see as an integral characteristic of their national spirit. I cannot speak for the other names, as I am less familiar with them.
Borges's Tlön, Uqbar, Orbis Tertius uses faraway places (faraway, of course, from his readers' tacitly presumed location) and "exotic" cultures to build portals into a new world. Would an Iraqi be offended, bewildered or amused to find Borges's portals constructed in her backyard at the dead of the night? I have never seen a reaction video, but I believe the last few wars were fought over different issues. And if we are to preemptively declare any inhabited regions of the earth off-limits for construction, where should Borges build his portals? Antarctica?
In other news, upon a suggestion of an alternative provenance of the name, Jeremy Brazas has admitted that
Reasons why “Hawaiian” should be kept don’t hold up and I’m ready to move on from it.
I'm wondering if the m.se editor will also be convinced.
Rolling back edits
If an edit does not improve an answer, does this automatically mean that a rollback is appropriate? When the author does the rolling-back, the answer is clearly "yes", and the same holds when the edit tangibly worsens the answer (e.g., introducing errors). The case at hand is less clear-cut; the politics aside, it's not much different from an overzealous American editor "fixing" British spelling. If the author doesn't visibly mind, should we take the trouble (and, in the case of mass edits, flood the front page again)? I don't have a good answer or even suggestion here; this is up to the community to decide.