I also have some thoughts!
So, there is a community of people who are well versed in creating “web apps”. They live and breathe this kind of thing. To them, having an editor that’s a web app means they instantly feel at home, with a cozy feeling. They also know how to hack it.
Another aspect is that web apps do make certain things pretty easy, such as styling via css. For example, the cursor is a DOM element and you can style it via css. Instead of blinking, you can make it pulse, by using css animations. If you build an app a different way, then you have to program such things in. Here, you can edit your personal style file to change the cursor appearance.
Now you can say who needs pulsing cursors (I don’t), I’m just using the example to show how text editors and presentation are not that far apart.
Because it’s easy to display stuff in a web browser, it’s also easy to display such stuff in Atom. For example, there is a markdown preview package for Atom that just compiles your markdown source code into HTML and then displays that. Little code required.
Do you need such fancy stuff in a text editor? Well, consider the preview-latex package for Emacs: it converts LaTeX source code for formulas into images of the formulas and lets you see the images instead of the source code. When the cursor moves into an image, it is “exploded” and you see the source code instead. I’d say that makes editing a LaTeX document with lots of formulas much easier. Atom can do the same. Of course, the user community is different, so I’m not sure might have that itch to scratch it.
How about on-the-fly rendering of Javadoc or JSdoc comments so that you can actually read them easily while browsing the code?
You mentioned that UE can be extended, but the extension is not too deep. That’s the point. In Atom, extensibility goes very deep, right into the core, and then out the other side. I also looked at Sublime. It’s got nice extensibility in Python (which is a nice language), but it also just scratches the surface.