unicode decode localizations

[dsschult]

Currently any unicode -> ascii conversions are region-independent and directly convert a symbol the same worldwide.

Some known issues:

• German umlauts
• Japanese kanji

Options:

• detect locale from institution and apply either unihandecode or manual replacements
• allow user customization of ascii rendering

The biggest issue here is username selection, which must be ascii. #9 would solve this.
Other issues are potentially in sorting users, or with adjacent applications (authorlist, mou).