Travis currently likes to fail because the tests do not output anything

for 10 minutes, more often with Python 3.2 than with others. This could be
a problem with our test code, but just as likely a problem with Travis,
so we'll try whether a travis_wait command helps.

.travis.yml

@@ -11,6 +11,6 @@ before_script:
     - python manage.py check
     - python manage.py validate
 script:
-    - coverage run manage.py test
+    - travis_wait coverage run manage.py test
 after_success:
     - coveralls