I've never seen any system like that. It always distinguishes between action and speech in what I've played, so you'd see something like:
At the bar, Bob looks at John with narrowed eyes. "Y... t me, pu..?" He yelled, "G...O..! Now, ... .... .... ch...."
Or whatever. And if people really wanted to yell, they wouldn't use tabletalk for it, usually.
Anyway, reading your layout of rationales, I think that's very well said.
It's very interesting to me that you would distinguish immersion and story. I've never thought of them as very separate concepts before, but now that I see that, it's enlightening. They're not the same thing. Most systems tend to prioritize one over the other to some degree, even while trying to have both, like being bilingual -- there's going to a first language in most cases.