Utf8StringBuffer incorrectly handles characters outside of the basic multilingual plane

In the Apache Solr project, we had some problems with jetty supporting Unicode characters outside
of the basic multilingual plane (codepoints greater than 0xffff).

The problem boils down to Utf8StringBuffer, all of the utf8 logic looks generally correct, but at
the end when it is time to append to the underlying StringBuffer, it needs to append the two
UTF-16 surrogates for characters in this range... its a rather simple fix.

The patch is against http://svn.codehaus.org/jetty/jetty/branches/jetty-6.1 (I hope this is correct?)