CFormatPart should probably be generic to allow String or Vec<u8>

I'm not sure that is necessary, I am using String as Vec here for bytes. As String stores char as utf-8 so I don't think any cost here.

@coolreader18 do you have an example which distinguish bytes and str for the case?

Well bytes can have non-ascii and still be formatted, while a String wouldn't conserve that:

>>> b'\xaa %b' % b'123'
b'\xaa 123'

Specifically, the u8 as char casting in this PR would map the non-ascii byte \xaa to the utf8 character sequence 'ª' == b'\xc2\xaa'. It's probably better than panicking, but it's lossy when it shouldn't need to be.

@coolreader18 do you have an example which distinguish bytes and str for the case?