Also don't skip mapping of glyphs which are empty, if the corresponding charCode is included in toUnicode.