Make Language::formatNum() handle TB through YB
Rewrote code to be simpler and less indenty
Though, something like formatBitrate might be be better in future... We'll see!
* Add a $noAbbrevs parameter that causes the 'seconds', 'minutes', etc. messages to be used instead of the 'seconds-abbrev', 'minutes-abbrev', etc. messages
* Add the 'seconds', 'minutes', 'hours' and 'days' messages
* Change the -abbrev messages to take a parameter rather than having the number prepended to them. This is for compatibility with 'seconds' et al, which need the parameter for {{PLURAL:}}. It also generally makes more sense. This does BREAK the messages in non-English languages that override them; Niklas told me to leave this alone and ping the TranslateWiki folks
* Introduce an 'ago' message for '$1 ago'. Not currently used in core, but I want to use it in an extension and it seemed stupid not to have such a thing in core.
* Refactor the function to use message objects and pass the number as a parameter
* Add tests! They exposed a subtle bug in my first iteration; all hail tests!
* Some fixes/changes to truncateHTML() based on tests
** Something like "<span>hello</span>" ends up as "<span>...</span>" instead of just "..." for relevant cases)
** If we get something like "<span></span" in, just return it back instead of ""
* Renamed $dispLength -> $dispLen in truncateHTML()
* Reject underscore in validation
* Still case unsensitive
* Corrected tests using underscore
Follow up r83160 which was a follow up of r82927 (language code validation)
A language code may contains the underscore character (be_tarask)
and might as well be upper case (FR).
Add tests for Language::isValidBuiltInCode() against some language codes
