Fix bug with counting utf-8 character lengths.

[benburkhart1]

This resolves the issue in:

#46

[nabeards]

This also appears to resolve my issues in #50 !

[boxerab]

LGTM !!

[micolous]

While it may fix the issue, this change duplicates _utf8_newline, and rolls back increments in the for loop.

I'd propose something much simpler:

size_t utf8_line_length(const utf8_char_t* data)
{
    size_t n, len = 0;

    while (0 != data[len]) {
        if (0 < (n = _utf8_newline(data + len))) {
            return len + n;
        }

        len += utf8_char_length(data + len);
    }

    return len;
}