Search &[u8] haystacks in regex_lite

[01mf02]

To quote the regex_lite docs:

Currently, this crate only supports searching &str. It does not have APIs for searching &[u8] haystacks, although it is planned to add these in the future if there’s demand.

This PR adds such an API in form of regex_lite::bytes::Regex, mirroring regex::bytes::Regex.
Thanks, @BurntSushi, for leaving some breadcrumbs in the source code that made this easier, in particular your hint in interpolate::bytes. This is much appreciated, as well as all your effort that went not only in the careful and beautiful API design, but also your very detailed and user-friendly documentation. Thanks a lot for your awesome work.

I made this PR because I need this functionality as part of jaq, where I want to transition from String to (a variation of) Vec<u8>, inspired by your "UTF-8 by convention" idea.
I did not cargo fmt this PR in order to make reviewing as easy as possible.

(Note to self: If I ever have to convert a lot of b"foo" to &b"foo"[..] again, this is how to do it in vim for all matches on the current line: :s/b"\([^"]*\)"/\&b"\1"[..]/g.)