I agree in part, and disagree in part. I like that gemini has a "native" markup format, and that its simple and bare-bones as it is. It's a communications baseline, and other things are negotiable between client and browser?
I'm waiting for folks to look at gemini, see only part of what they want, and make markdown-native web protocol, a wiki protocol, basically. I think that would be very cool, and it is what drew me to gemini before I figured out it was really super-gopher. Like what if there was a mediawiki protocol? You'd have your tables! Anyway, there's no one stopping anybody for trying to make that happen. The true interwiki promised in the days of c2, could actually happen!... if some people want to make it happen.
I'd like to say again that by competing with the modern web to do things that gemini isn't good at, gemini gets to lose those contests, and this should limit the success of efforts to to make dancing bears out of gemini.
So you say the "effort". There are 50-some line gemini clients, and more than a few people have coded graphical clients in a few days by themselves. Same with servers. Is modifying firefox or palemoon really going to be a "better spent" effort?
I think screen readers can read text pretty well, also, since you've declared that to be a goal for some reason.
The thing is, when the web grew up, there was no not-web. Gemini occupies a swim lane, and if you want more, even a gemini page can link to a web page to do the heavy lifting of the modern web. So aside from making the bear dance, there's little interesting in forcing inline images. It will be not that the bear dances well, but that it dances at all, right? There is definitely a part of the Gemini community--from what I've seen--that is really excited to see the http 0.9 grow up that they missed the first time. But so far cooler heads are using their persuasion to knock most of that urge down. What will save gemini from turning into the modern web is that the modern web will be right there the whole time. I guess the perverse server and client could collude to deform the gemini protocol to look like http, but why? "=> http://www.mysite/recipe-with-inline-images/ Step by step recipes " in your .gmi file would do this so much better.