ByteSlice should offer more functions taking u8

I'm using bstr to help parse a few formats, some of them are text (but not necessarily utf8) and some of them are binary but contain strings.

I have a few things that I wished were present:

• The various split functions should have variants for passing a byte instead of just str/char. I ended up using split_str(b"\0") and split_str(b"\xff") a few times which is going to be less efficient than directly invoking memchr.

• Versions of fields_with/trim_start_with/trim_end_with which pass their function the byte instead, and don't bother with UTF-8 decoding.

It seems possible that you're more interested in this being useful for probably-text case than for (e.g. the emphasis is on the str, and not the b). If that's the case, sorry for this and the next bug I'm going to file!

Start by locating the existing split functions and the fields_with, trim_start_with, and trim_end_with implementations in the bstr source. Compare their current string- and character-based behavior, then determine the byte-oriented variants and verify that they avoid UTF-8 decoding while preserving the existing APIs and semantics.