it's not wrong - but it's also not entirely correct - it's in between, depending
on which aspects of integers you focus on - Nim integers and C integers share
some semantics and don't share others. `int32` is "good enough" on windows
because it shares two important properties (size and alignment) and that's all
the nim compiler needs to figure things out.

perhaps an example will help :
 the intent of `clong` is stated as: "This is the same as the type `long` in
C."
 the intent of `int` in nim is "an integer of the same size as a pointer"
 your new documentation says "`int` on non-Windows."

now we add a platform where pointers are 128 bits. on such a platform, `clong`
might be 64 bits but nim int will certainly be 128 bits. it follows that at
least one of the 3 points is now false for the 64-bit clong case - to fix it, we
have to change something, and that something is your new documentation.

this is the problem of your PR: it introduces a bug in the documentation that we
know today we'll eventually have to fix .. it documents an implementation detail
where in reality it would be more useful it it was teaching users about this
potential future