Fixed a small bug. doctest didn't handle unicode docstrings containing
non-ascii characters.
diff --git a/Lib/test/test_doctest2.txt b/Lib/test/test_doctest2.txt
index 0d7d1d5..2e14856 100644
--- a/Lib/test/test_doctest2.txt
+++ b/Lib/test/test_doctest2.txt
@@ -5,3 +5,10 @@
   >>> import test.test_doctest
   >>> test.test_doctest.sillySetup
   True
+
+This test also has some (random) encoded (utf-8) unicode text:
+
+  ЉЊЈЁЂ
+
+This doesn't cause a problem in the tect surrounding the examples, but
+we include it here (in this test text file) to make sure. :)