I would consider this a test of the quality of the documentation, not of the implementation. It's important that the examples in the documentation are actually correct. For example, a doctest in Doc/library/datetime.rst passes an "hours=" keyword argument to datetime.time when "hour=" is correct. It's going to be hard to find such bugs in examples without an automated check like `make doctest`.

I'm open to simplifying the patches by removing testsetup blocks. For what it's worth though, the docs already contain 17 testsetup blocks.