> As far as the C++ standard is concerned, it is. More precisely, it
happens before dynamic initialization and running main(), i.e. before
any of the program's code runs.

Yes, and that is still run-time. Just because it happens before
dynamic initialization doesn't make it not run-time. The process of
running the executable and mapping memory is obviously run-time, not
compile-time. You're arbitrarily drawing a line before dynamic
initialization and saying that all the run-time work the operating
system does up to that point is not run-time ... because ...uhm?

> So, as far as the standard is concerned, the object could as well exist eternally.

No, it would not be valid. This would imply that the object's address
is also eternally a constant, but the address can be different each
time the program runs. The only thing that's guaranteed is that the
value it is initialized with AT RUN-TIME is constant.

> I disagree. I don't think that diluting the meaning of constinit is a good idea ...

That's understandable. This is the biggest hurdle for any such
proposal; I'm sure there would also be voices in the committee saying
more or less the same.