Encode to Vec<u8>

[rubdos]

Adds Message::encode_to_vec(&self) -> Vec<u8> and Message::encode_length_delimited_to_vec() -> Vec<u8> on std-enabled configurations.

This is one line shorter than writing

let mut buf = Vec::new();
msg.encode(&mut buf).unwrap();

and replaces it by

let buf = msg.encode_to_vec();

The main reason I do that is because it's confused me (and others) before when writing PROST code in combination with symmetric crypto code (where encrypt(&self, &mut [u8]) overwrites allocated bytes, instead of extending the buffer like PROST).

It also omits the extra capacity checking of encode (since Vec is pre-allocated because this patch uses Vec::with_capacity()).

Fixes #154 in spirit, if you agree that Into<Vec<u8>> for Message is a slightly worse idea (i.e., .into doesn't specify whether to have length encoding or not, so it might yield slightly more unreadable code).

EDIT: CI failure seems to be due to use of a new rustc feature? I don't think it's this patch.

[rubdos]

Rebased on master, subtle nod nod to the owners ;-)

[Jasperav]

I guess this fixes https://github.com/danburkert/prost/issues/461, what do you think @rubdos? Adding this change as an example in the readme.md will help newcomers to this library :)

[rubdos]

I agree that it'd fix #461 if it's added to the README. I'd have hoped to have some feedback from the maintainers though by now :/

[Jasperav]

@danburkert please review

[rubdos]

This is a relatively simple QoL patch; would any of the people that have commit rights please take a look?

[LucioFranco]

sorry for the delay on this!