Fix up documentation around no_std

[steveklabnik]

1. Fix the sections in the book to have the correct signatures. I've
   also marked them as ignore; there's no way to set the no_std
   feature for libc, so it pulls in the stdlib, so this wasn't even
   testing the actual thing it was testing. Better to just ignore.
2. Correcting libcore's docs for factual inaccuracy, and add a note
   about language items.

Fixes#33677

r? @alexcrichton

[alexcrichton]

I wonder if we can also make sure these is all plastered with "zomg this is unstable"? Documenting things like symbol names makes me a little queasy as I don't want anyone relying on that ever basically

[alexcrichton]

No traits here though?

[steveklabnik]

Whoops, this was just the previous text.

[steveklabnik]

I wonder if we can also make sure these is all plastered with "zomg this is unstable"?

So, the book part is explicitly in "unstable rust", but making it more clear in libcore's docs is something I can do!

[alexcrichton]

Ok! Sounds good to me, looks like travis failure is legit though:

run doc-book-no-stdlib [x86_64-unknown-linux-gnu]
running 2 tests
test _0 ... ignored
test _1 ... FAILED
failures:
---- _1 stdout ----
<anon>:7:1: 7:19 error: use of unstable library feature 'libc': use `libc` from crates.io (see issue #27783)
<anon>:7 extern crate libc;
^~~~~~~~~~~~~~~~~~
<anon>:7:1: 7:19 help: add #![feature(libc)] to the crate attributes to enable
error: aborting due to previous error(s)
thread '_1' panicked at 'Box<Any>', src/librustc/session/mod.rs:170
note: Run with `RUST_BACKTRACE=1` for a backtrace.
failures:
_1
test result: FAILED. 0 passed; 1 failed; 1 ignored; 0 measured

r=me other than that though

[steveklabnik]

Whoops, I was checking as I wrote it, but missed it in the end. Should be good now, let's see what travis says.

[alexcrichton]

@bors: r+ fcecdac

[GuillaumeGomez]

@bors: rollup