Recent updates

[Curt Arnold]

I done yet another iteration of character encoding.  The structure is 
still based on the JDK 1.4 CharsetDecoder and CharsetEncoder, but now 
most of the heavy lifting is done with a small set of implementations 
within log4cxx.  apr_xlate is used as a fallback in case the nature of 
wchar_t is not obvious.  apr-iconv is no longer needed and is not built 
as part of log4cxx's build process.  I'm sure that there will need to 
be some tweaks to support some platforms (usually would be indicated by 
some missing definitions during compilation or linking).  Non-windows 
platforms where __STDC_ISO_10646__ is not set will likely fail due to 
missing functions in UnicodeHelper.  For those platforms, either iconv 
needs to be available or functions that convert between UCS-4 and 
wchar_t values need to be provided.  At some later time, I'll review 
the apr-iconv source and see if I can see how they anticipated those 
platforms.  Eventually, will probably provide a build option to 
hard-code the external encoding to UTF-8 which would allow all the 
encoding/decoding steps to be short-circuited with copies when you know 
at compile time that would be safe.

I've integrated a patch submitted by Andreas Fester which contained 
reworks for the TelnetAppender and supporting classes implemented with 
APR's socket implementation.  I haven't tested it and had to hack some 
things to get it to compile on Windows.  If you have any comments or 
issues with it, please let the list know or add a comment to 
http://issues.apache.org/jira/browse/LOGCXX-80

My next task will be to jump over to log4j to address and issue with 
the RollingFileAppender's to get them ready to be ported over to 
log4cxx.