Let's not forget other more recent stuff that still happened before AtomOS, like Redox. I seem to recall seeing a couple others recently as well, but for some reason I'm drawing a blank now. I think the big deal about this is mostly just that it's an effort outside of Microsoft to do this, and compile it to machine code rather than just use some kind of "higher order" OS that needs to run on top of something else -- and even there it's not unique, even if it looks more interesting than FlingOS (which is evidently just a new take on the "toy kernel/OS for teaching" idea).
I thought of writing an operating system in C++ in the 1990s. I didn't achieve it until after the turn of the century. But I definitely thought of it.
The "everyone" who knew these things about C++ was a small set of vocal people largely from the Unix and C worlds. Other people who were at the time developing Windows NT device drivers in C++, to pick one example, knew that a lot of the received, and loudly recirculated, wisdom about systems programming and C++ was just bunkum.
What I was trying to convey with that comment was that C++ was looked down for systems programming, it required Symbian, BeOS, OS X IO Kit to change that mentality.
Even Windows that you mention, which always was a more welcoming place for C++ tooling, Microsoft only started to officially support it with Visual C++ 2012 when the /kernel flag was introduced.
So even that microkernel could have been eventually written in System C#, if there was enough management support to keep developing Midori and improving the language to the point C++ wouldn't be needed.
Instead the project was canceled and since in IT, seeing is believing, there will be idea for many, that there wasn't any other way to accomplish it.
I love the way Alan Kay and others like him think, it is not what the technology give us today, but what it can give us tomorrow in a few years time.