No longer run a safe HTML string through the HTML purifier (since it's already marked as safe).
This also addresses the issue of calling purify() when no purifier is installed. In that case, we'd run clean() on a clean string (double HTML encoding).
If this approach doesn't work out, we can still modify the fallback code of purify() to check if the string is already clean before calling clean() instead of purify().
Refactoring of Xss_Security_Test.
t() and t2() return a SafeString instance.
TODO:
- Update all code to use SafeString where appropriate.
- Update golden fole of Xss_Security_Test
- Stop reporting CLEAN vars in Xss_Security_Test
