Document platform-specific differences for std::process::exit.

[frewsxcv]

Fixes #35046.

[rust-highfive]

Thanks for the pull request, and welcome! The Rust team is excited to review your changes, and you should hear from @alexcrichton (or someone else) soon.

If any changes to this PR are deemed necessary, please add them as extra commits. This ensures that the reviewer can see what has changed since they last reviewed the code. Due to the way GitHub handles out-of-date commits, this should also make it reasonably obvious what issues have or haven't been addressed. Large or tricky changes may require several passes of review and changes.

Please see the contribution instructions for more information.

[frewsxcv]

I was going to add a sentence at the end like:

For specific information about your platform, please check <insert resource here>

But I wasn't sure what resource to put there. man 3 exit? "your systems 'exit' function"?

¯\_(ツ)_/¯

[alexcrichton]

I tend to like to cite the opengroup docs as they've always seemed "extra official". Maybe a link to that could be added here?

[frewsxcv]

It looks like your opengroup link says exit only takes into consideration the least significant eight bits. I have hesitations about just citing that as it doesn't match the behavior on Windows which takes into account all 32 bits with no truncation (as per this comment by @retep998).

[alexcrichton]

Oh sorry to be clear that opengroup page is a Unix-only thing, which is basically just explaining why only 8 bits are transmitted to the parent on Unix.

[nagisa]

Windows is the only tier 1~2 platform which I’m sure passes through all the 32 bits. Every other target may be passing through whatever number of bits they want, as long as its not less than 8 bits. I’m also not sure it is worth mentioning platforms which have obvious behaviour at all (i.e. Windows). Since tiers are an always-moving list, I’d perhaps not tie any explanation to that, but I still feel like it could be worded better. For example:

## Platform-specific behaviour

**Unix**: On UNIX-like (or POSIX, whichever you prefer) platforms it is unlikely that all the 32 bits of 
status code will be visible to a parent process inspecting the exit code. On most UNIX-like platforms 
only 8 least-significant bits are considered/evaluated/whatever is the best word here.

---

I do not think it is necessary to link to any external references here. I’m confident that mentioning this exceptional behaviour is necessary, but in case of references you’ll either end up having to reference documentation for every single supported platform (some of them have no documentation at all) or linking a POSIX spec, which is useless if you’re trying to figure out exact behaviour on your particular platform.

[frewsxcv]

Great feedback, thanks! I incorporated in your block of text you wrote above and made some minor tweaks.

[alexcrichton]

Thanks @frewsxcv! In retrospect we should have probably made this argument a u8, but alas.

[nagisa]

This is a great example.

[alexcrichton]

@bors: r+

[bors]

📌 Commit 0971cb1 has been approved by alexcrichton

[frewsxcv]

@bors rollup

[frewsxcv]

This is going to fail on Windows, because it's a non-zero return code. So I'm going to mark as no_run.

[frewsxcv]

@bors r=alexcrichton

[bors]

📌 Commit 4d392d3 has been approved by alexcrichton