Remove mbrtowc(), and use utf8_char_len() instead.

Remove mbrtowc(), and use utf8_char_len() instead.

- In get_counts(), revert the line_char variable to its original
  behavior: keeping track of the character count, not the byte offset.

- Introduce a new parameter to get_counts() to keep track of the byte
  offset.

- Use the byte offset data in Source_file::quote_input() to make sure
  the source line is split at the correct place.