Support more than UTF-8/16 in word metadata import

The meta-data import code was only considering two possible codepages:
UTF-8 and UTF-16.
Word documents tend to use local encoding, so while this behaviour seemed
flawless with US/UK documents, completely different encodings were broken.
Instead of considering only UTF-8 and UTF-16, use QTextCodec and try to
handle as many encoding as possible that way, warning if they are not
found.

See https://bugs.kde.org/show_bug.cgi?id=406014 for example document

BUG: 406014