If a majority decides to fork leaving any legitimate minority excluded, then the minority is screwed

You have no say over this, as much as you might like to. If people want to move to a different chain (including a chain which forked from a previous chain), with some differences in state but some commonalities, they can absolutely do so. There's nothing really to discuss here, regardless of what you or think. No mechanism is possible that can prevent it. If you think it is a bad idea, your only real option is to not use it.

The old chain is lost to history (except in a couple of very famous, high profile cases where the minority has spent millions trying to keep their "forks" alive).

In the case of proof-of-work chains, yes it is true that mining costs money. Whether this is any sort of deliberate "spending" of millions for a political agenda or just the usual case of miners chasing rewards is somewhat a matter of opinion but the latter certainly does happen (as when relative profitability of the chains shift and miners follow it). But in any case this does not apply to PoS chains which do not have mining cost. It is arguably a version of the famous nothing-at-stake problem that infinitely many forks can coexist with no added cost. The only thing that really determines which ones do is whether people find them attractive to use.

Nobody in their right mind would want to be on either chain!

People have repeatedy made these sorts of claims about various hard forks and chain splits in cryptocurrency history and always been wrong. It turns out that it is almost impossible to come up with anything (even the most incredibly stupid ideas, in my opinion, or probably yours), which "nobody" would want. Perhaps this simply means there are a great many people not in their right minds, but again what you or I think does not matter.

If Ned were to refuse to hand over GitHub access to the community, then there would be justification for forking and leaving him out

This has been requested repeatedly by witnesses and has been both explicitly and implicitly (by failing to take actions toward it) rejected. Apart from actual handing over, there has been repeated resistance to even allowing: a) third parties to perform some of the development work, to help develop a less centralized process and competence; b) including non-Steemit developers in planning, design, and roadmap processes; c) accepting work from non-Steemit developers which disagrees with the Steemit-determined roadmap. This included threats of using the Steemit stake to vote out witnesses in favor of it. So there we are. Draw your own conclusions I guess.