str api should offer a method to convert a `&u8` slice into an `&str` with same lifetime

[nikomatsakis]

Since strs and vecs share the same representation, there should be a way to "cast" a &u8 slice into an &str with the same lifetime.

There are a few FIXMEs in tydecode where this could save a copy.

[sanxiyn]

Isn't this an unsafe operation? Not all &u8 are valid &str.

[nikomatsakis]

I presumed there'd be an assertion as well as an unsafe variant, as we have for str::from_bytes

[Kimundi]

@nikomatsakis Could you confirm the PR request did add what was wanted in this issue? I remembered that there where some difficulties to make it work because a &str slice has to cover one byte more than a &[].

[nikomatsakis]

Looks good, thanks.