I understood it as Windows developers had to manually deal with segment limitations since Windows supported running on pre-286 CPUs without protected mode (Wikipedia says Windows 1-3 all supported the 8088). OS/2 just made the 286 a minimum requirement so they could rely on a CPU with more modern features.

The 68k didn't come with an MMU like the 286 so MacOS couldn't rely on virtual memory like OS/2 did but at least the flat memory space meant you didn't have to juggle 64k segments

Yes, outwith the idea of Family API programs (which couldn't use Presentation Manager and whatnot anyway) OS/2 1.x did target the 286 as a minimum. But that doesn't mean that DOS+Windows didn't use the features.

It did. It was bi-modal. There were at one point switches to the WIN command to tell it whether to come up in real mode or 286 protected mode. In the latter it definitely did use the features of protected mode.

It was the bi-modal nature that was the problem. Essentially, they had to design a whole layer that simulated when in real mode all of the load-on-demand stuff that the processor architecture supplied for free in 286 protected mode, and make it so that the thing would all work either way with no changes to applications.