Fwiw, I moved to Kbin.social more than a year ago at the time of the Rexodus, then bounced around multiple Lemmy instances, and this is the first that I have heard of that community iirc. And similarly I had never heard of this !newtolemmy@lemmy.ca either until you told me about it yesterday:-). The former is listed on the Lemmy.ml main communities page, but it is #43 (by the default sorting method - it looks like Users/Month?) so quite buried (and likewise this community doesn't even make the first page, even locally for the lemmy.ca instance it is on).
Anyway, it's a really good point that if this were to be taken forward, it should be noted that Lemmy.ml is one of the oldest instances, and that the admins/maintainers are the Lemmy developers, so anyone involved in advancing the Fediverse forward, or perhaps keeping abreast of the updates for the Lemmy code, may want to not block it. i.e. it is good to list both the costs & benefits of doing or not doing that.
That said, (a) I don't think those are scenarios for the mainstream public - e.g. people fleeing X b/c it is becoming too politically extreme (how ironic then that we are even more so, not so much in an average sense but in a maximum one, though just calling it by a different name:-P) - and (b) anyway it would be good to have such things be opt-in, rather than have to opt-out of them by default. So yeah, somehow finding a way to explain the situation clearly and cleanly, and without so many words as I have used here. Plus, how would people even find this information? e.g. going to https://lemmy.world/, you see e.g. a "getting started guide", but there are already so many comments on it that the auto-scroll takes many times to populate them - and using the web UI at least, we can't really search through comments, until they are all loaded.
So maybe something could be done on the Lemmy Explorer? There is already an option to "include suspicious", maybe something similar could be done to "include politically extremist"? Although I suspect in that case it would end up going back to being opt-out again rather than opt-in, though at least it would move things forward in the tiniest manner. Otherwise, for good or ill, Lemmy remains relegated to basically users of linux who will put up with such things, whereas mainstream users will simply not bother.
But unlike Reddit, the communities here change very frequently - new ones spring up all the time, and old ones receive fewer content delivery, or sometimes new ones spring up out of an old one.
Though if communities could be trusted to label themselves with category labels, that would allow them to dynamically update, moment to moment as someone was in the mood to e.g. take a break from politics after reading that for an hour, and now wanting to relax with e.g. non-political memes.
It would get complicated to label them, e.g. !memes@lemmy.ml is anything but non-political, and despite the leftist stance of midwest.social, the community !lotrmemes@midwest.social is mostly devoid of politics. So like... are such labels up to whatever the user wants, or whatever the community mod does, or an instance admin...? It would depend on the implementation I suppose.
As it is now, smaller communities tend to get lost even in the Subscribed feed - e.g. the largest poetry community is !poetry@lemmy.world with "only" 1k subscribers - so having multiple feed categories to switch among may allow less-populated communities to flourish more readily.:-)