Issue: Usernames are Not Normalized

This is particularly harmful during imports - but can cause display issues in some other cases.

If the user registered to the forum with the registration form using a different charset than VaultWiki, then the encoding of the username cannot be trusted. It is particularly noticeable when doing user searches from wiki forms (add moderator, contributions, etc). If the username contains a special character, the database would contain a different byte code than the form is requested, since the username was created via a different charset. This results in the requested user not being found.

When importing, this means that many records that should be attributed to a user would instead be attributed to a guest that happens to have the same username.

There is not a way to guarantee that the user will always be found, because particularly in vBulletin, the admins can create many languages on their sites that use many different charsets, and the creation language is not stored with the username (even though that would be near useless during a reverse lookup anyway - re: the reason VaultWiki went full UTF-8). Thus, we must assume for vBulletin that all users registered to the forum while viewing the registration form using the forum's DEFAULT language option.

In this case, we can lookup the default charset for vBulletin and convert usernames before doing reverse lookups or importing data. This will increase the success rate for lookups involving usernames with special characters.