Add equivalents for INADDR_ANY, INADDR_LOOPBACK to std::net

[luser]

POSIX defines:

• INADDR_ANY, which is 0.0.0.0 as an in_addr
• INADDR_BROADCAST, which is 255.255.255.255 as an in_addr
• in6addr_any + IN6ADDR_ANY_INIT, which is the same concept as an in6_addr (with wonky variable + macro usage because structs in C are terrible)
• in6addr_loopback + IN6ADDR_LOOPBACK_INIT, which is ::1 as an in6_addr

glibc additionally defines:

• INADDR_LOOPBACK, which is 127.0.0.1 (I don't know why this isn't in POSIX)
• INADDR_NONE, which is just used as an error return, so not really useful in Rust

I think it would be useful to have at least consts for the ANY + LOOPBACK values for IPv4 and IPv6 in std::net.

From cursory googling while looking up the sources above I did find a few people wondering why these were missing from the Rust standard library:
https://users.rust-lang.org/t/where-is-inaddr-any-for-std-net/2337
https://www.reddit.com/r/rust/comments/3gi34h/inaddr_any_in_rust/

[luser]

Currently in Rust you either write these out as strings (which all the examples in std::net seem to do), like TcpStream::connect("127.0.0.1:12345"), or you call Ipv4Addr::new, like TcpStream::connect((Ipv4Addr::new(127, 0, 0, 1), 12345)). They're both a little clunky, and the former requires parsing a string to get a value that is always going to be the same.

[Mark-Simulacrum]

cc @rust-lang/libs, I assume we probably want this developed out-of-tree as per usual? Probably in the net2 crate?

[sfackler]

These are just a couple of new consts, right? I don't see any reason to not just add them (unstable at first of course).

[Mark-Simulacrum]

I believe so, though I'm uncertain. @luser Are these just a set of constants?

[luser]

Yes, these are all just constants.

[dtolnay]

I believe we are ready to consider a PR exposing these constants.

[luser]

This was fixed in #44582

https://github.com/rust-lang/rust/blob/master/RELEASES.md#stabilized-apis

Hooray!