"what am I missing"

I guess you underestimate a bit the distinction between the roles of user and of admin/operator of an instance and the different legal constraints that go with both.

Whatever gets "published" in the Public feed on an instance needs to conform to the legal standards of the country the servers of the site reside. May that be laws of libel, of controversial content, or other possibly criminal activities. At least in principle, the operator of the instance can be held legally responsible for the content appearing on "his" or "her" site. (Whether an instance should or could even be seen as some possession or "real estate" is a question I'd rather set aside for now.)

As far as I know these conditions of legal responsibility do not necessarily apply to the content that appears in the Whole Network feed via federation. Here the laws of the servers' country as well as the orginial licensing conditions of the federating site "from where" the content stems seem to apply. But when federated posts from various sites mix in conversation threads this can create the situation that e.g. in an abuse situation (across instance) the addressee of the abuse cannot properly counter should his or her counter post violate rules of his/her instance or even the legal conditions of the ocuntry (far fetched, admittedly). Think also federated bots, spam, unwanted content as further examples. In such cases and cirumstances blocking and muting appear to many a practical solution.

To me instance blocking is a borderline case to which I came to grudgingly agree even though it undercuts federation and creates bubbles of sub-diverses. But individual blocking and muting seem viable and reasonable to me. Of course this is no answer to the more foundational questions you raise. But on a practical level, I can see their benefits.