Add support for unicode width

[ratmice]

In just testing some byte offset to char index conversion code, I noticed
#41

When testing with "🦀".
Naively just modifying char_width() prints multiple instances of the unicode character in question.
"🦀" becomes "🦀🦀",

I left this behavior enabled for whitespace since I believe it is a part of the treatment of '\t' tab characters.

This perhaps doesn't fully fix the issue as reported, they might need a boolean to Config setting, which sets things up to call either width and width_cjk, but I don't know the right behavior to shoot for regarding whitespace in cjk.

[zesterer]

Thanks, this seems like it's am improvement on master's behaviour if nothing else!