Normalize Unicode input to normalization form C. Most of the time input

Normalize Unicode input to normalization form C. Most of the time input
is already in this form and it shouldn't take very long to verify it.
There is still optimization to be done though.

Partial fix for http://bugzilla.wikipedia.org/show_bug.cgi?id=240

Will also need to verify correct UTF-8 sequences and strip characters
that are illegal in XML.

Some input may not be going through this verification yet (eg the uploaded filenames)