[PATCH] Avoid character remapping if font contains the same character multiple times

When using font Meiryo for japanese PDF output, some japanese characters where remapped, e.g. 青 (decimal unicode: 38738) was remapped to ⻘ (decimal unicode: 11992). My patch contains a changed MultiByteFont.findCharacterFromGlyphIndex() method that tries to preserve the original character, if possible, but falls back to the current behavior if not.

Preserving the original characters is important for a more predictable search behavior in PDF viewers. Since the characters were remapped, searching for the original character didn't show any results.