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.
diff --git a/Lib/encodings/utf_32.py b/Lib/encodings/utf_32.py
index 622f84b..4bbd22a 100644
--- a/Lib/encodings/utf_32.py
+++ b/Lib/encodings/utf_32.py
@@ -125,7 +125,7 @@
         if byteorder == -1:
             self.decode = codecs.utf_32_le_decode
         elif byteorder == 1:
-            self.decode = codecs.utf_32_le_decode
+            self.decode = codecs.utf_32_be_decode
         elif consumed>=4:
             raise UnicodeError,"UTF-32 stream does not start with BOM"
         return (object, consumed)