These things meant that later the Atari community and Atari themselves were able to extend the OS in a proper multitasking almost Unix-like direction (MiNT and MultiGEM) and bring it to new hardware, and new display formats and architectures, etc. Provided the applications being run were cleanly written (well, that's a big caveat...). For example -- we can now run TOS/GEM on an Amiga, that's pretty neat (though totally pointless).
In true Tramiel fashion they shipped the cheapest simplest thing they could. But it was something they were able to iterate on -- unfortunately they just did this too slowly. As others have pointed out, the Amiga had amazing hardware from the go, but its architecture became somewhat tied to that original hardware. It was more like a video games console than a workstation. They had a few years headstart on everyone else and then having (then dated) specialized hardware became a liability, not an advantage.
I built MicroGNUEmacs as a desk accessory so that I could run it at the same time as other programs.
I never saw uemacs compiled as a DA. That would have been a nice trick. The only microemacs I used on my ST was not a windowed application, was console only.
I chose the MicroGNU variant as it did parenthesis matching better than the alternatives. I ran it alongside Franz Lisp that I had also made into a GEM application.