path: Refactor set_extension and contains_nul

[Ryman]

Also some minor cleanup in Path related to this.

[alexcrichton]

Could you elaborate a little more on what the use case is for requiring the Container trait? I would expect that it's just "one more trait" to implement, and it doesn't have a particularly compelling reason to require.

[Ryman]

It was more that it felt implied, but I guess you're right and it could be restrictive/bothersome to require it.

Would you instead accept if I instead extend BytesContainer to have byte_len()? The default trait implementation should actually handle most needs then.

[alexcrichton]

I would figure that byte_len would be the equivalent of container_as_bytes().byte_len().

I think that this trait may be seeing a bit more prime time than it was intended for, sadly, it's mostly just to make Path::new "just work" most of the time.

[Ryman]

Updated commits do as mentioned above, feel free to close pr if unwanted.

[alexcrichton]

Would you be ok with omitting the addition of the byte_len() method, but keeping your refactoring of the path module? It seems like you had some good refactorings in there, and it'd be a shame to lose them!

[alexcrichton]

Stylistically, this if/else should look like:

ifcontains_nul(s){None}else{Some(...)}

[Ryman]

Nuked byte_len() addition and fixed if style.