Normalize reading keys to match expressions

[stephenmk]

Sometimes the usage of hiragana or katakana in reading fields in the JMdict database is not in alignment with the corresponding expression fields.

For example, the expression タチの悪い has a corresponding reading equal to たちのわるい in JMdict.

When headwords like this are displayed in Yomitan, the タチ part of the word receives たち furigana.

I suppose this is nice if you can't read katakana, but I'm not sure if that's a big concern. I think it might look better if タチ did not display any furigana here.

[stephenmk]

Fixed in the next release of Jitendex