That's not technically true, you could pass 999 to get the 9th subplot of 81, but I guess we really don't want to encourage this form too much anyway.

You're right, in both the observation and the conclusion 😄.

This type is not documented; should it be?

It's not documented as type, but in the description:

            In rare circumstances, `.add_subplot` may be called with a single
            argument, a subplot axes instance already created in the
            present figure but not in the figure's list of axes.

I'm not even sure what this is good for or how to achieve that, but that code path triggers on the tests. I've intentionally omitted this form the types. I assume adding it to the types without big warnings would probably do more harm than it's worth, because people will start to put axes from other figures in.