Fix stupid typo in Lib/encodings/utf_32.py which led to failing tests

on big endian machines.

Update documentation: UTF-32 codecs will be in 2.6.