If the journal demands you remove the appendix or supplementary, just remove it from the published manuscript. Then add it to your ArXiv submission.
1. The problem in math is not that the way the "orthodoxy" insists on presenting things in a suboptimal way, but that if the reviewers find a good explanation of your result, they'll recommend that you publish in a lower-ranked venue than your result would ordinarily merit (at least unless you solved a famous problem). So researchers are incentivized to make the presentations of the proofs as opaque as possible. You can see this in conference and workshop talks (which tend to happen pre-publication in math), in many fields speakers always avoid presenting _any_ proofs. Putting better explanations in an appendix, which the referees can read, simply wouldn't help.
2. Being easy-to-understand at first is actively punished, but even being easy-to-understand in the long run is not rewarded. You can always write an explanatory blog post after the publication decision has been made, but you won't put effort into writing one if you don't gain anything from it. This applies even more to the people in the example, who wrote on a typewriter in the 1970s. There were no blogs, and it was much harder to get an appendix through because page limits were physical limits, and the act of writing was much more onerous before the age of computers and LaTeX. There was no point to doing it given that it was actively discouraged.
That said, to present in "high impact" journals, you do have to write your results in some grand fashion where it is the greatest thing since sliced bread. But the results, not the proofs. Then again, the difficult theorem proving papers are rarely published in high impact journals.
Also, I have seen quite a number of papers where the arxiv submission is more updated/expanded than the published version. If you have some new framework that you want people to adopt, then it is in your benefit that grad students actually understand the ins and outs of it, so people so inclined do put in some effort to make their results accessible.
But reading this comment chain, am I correct to understand that this problem stems from the very essence of math itself? The people who live and breathe math just fucking hate sharing their passion with others?
What the hell.
Even if you solve a well known problem. I gave a talk once where I did so as a very junior student. During the Q&A a very well known senior person got up and basically asked me what's the point and this is all trivial anyway. Thankfully I remembered the exact place where the founder of the entire field had said just recently this is one of the hardest problems in the space and the problem he had hoped to maybe one day get to when he started the entire enterprise. The audience laughed and the guy apologized.
But I learned my lesson. Talks and papers need a little magic. For some people you can't just solve a cool problem, they need to think that they couldn't have done so and that they don't quite get how you did it. I now include something in every talk that I don't expect the audience to get just so what I'm doing seems "hard" to people who think this way.
Smart people can be incredibly biased and ideological. Some careers for smart people are even all about reasoning backwards from a given conclusion.
[1] redacted footnote
When I did that, a version of the derogatory sentence "The proofs are easy / non-technical." would appear in the reviews. Every. Single. Time. Of course, I have some independent confirmation that the proofs weren't easier than in any of my other work (e.g. my coauthors and I had to work just as hard to get them), but this led to having to resubmit them to less prestigious journals than the ones which normally published my work.
I gave up on this approach, and realize now that the opposite is more likely to be rewarded: out of 18 eventually-published papers, I only managed to piss off the referees enough for a revise/resubmit decision once, and I really went out of my way to keep the proofs vague that time.
Of course, I had a largely unremarkable career in a somewhat niche subfield: I'm sure there are levels where (a), (b), and more importantly the sheer speed required to get a result out are bigger incentives. And from yet other fields, I occasionally hear rumors of people who master the art of opaque writing and "parallel construction" only to make it difficult for others to get ahead of them (hi שַשֶׁ!).