Simplify instantiation of &BStr

[Byron]

This issue was spun off from this comment which states problems with creating instances of &BStr if type-inference bails out. The latter is typical within Cow and as one side of assert_eq!().

The intuitive way of doing this to me was BStr::new("") or BStr::new("hello") or BStr::new(b"hi"), similar to what's possible with std::path::Path::new(…).

[BurntSushi]

Aye. Were there any other helper functions you defined? I think it would be helpful to see those and see if we can bring them into bstr.

[Byron]

I am not sure about that 😅, as for no particular reason other than some familiarity I used fn b(s: impl AsRef<[u8]>) -> &BStr { s.into() }.

[BurntSushi]

Ah yeah. I wasn't sure if you had helpers for things other than construction.