[no change needed] I think wecould use subtests to provide more informative context (basically what you've got in each # comment header) when a test fails? but this is already in keeping with the existing style of this file, so not a big deal. these seem to nicely encode the current behavior state of ourre implementation.

The comments provide context to the reader of the code (so you will know where to add new tests), not when a test fails. The traceback already has all necessary context.

subTest() has two functions:

• Provide more informative context when a test fails. This is especially useful when traceback does not identify the failed test (in a loop, with generated test data).
• Allows to continue execution when a test fails and collect information about other test failures in one run. This works only when tests are independent andsubTest() is well granulated.

It has a drawback -- the traceback is limited and can be less informative ifsubTest() is used in a helper.