Parse indented numpy docstring examples

[gselzer]

For reviewers

• I did not use AI
• I used AI and thoroughly reviewed every code/docs change

I've never dug into mkdocstrings before, so the LLM was useful for pinpointing the issue and suggesting a quick fix. Afterwards I slimmed down a rather verbose test 😅

Description of the change

This PR is designed to enable the parsing of numpy-style code examples that are nonuniformly indented. Sometimes authors (read: me) might want to indent a code example to highlight it amidst the larger section - but griffe doesn't pick those up.

This PR also includes a test that fails without the changes.

Relevant resources

This issue was spotted in pyapp-kit/scenex#89

[pawamoy]

Thanks for the PR @gselzer. Have you considered simply not indenting your >>> blocks and use https://github.com/pawamoy/markdown-pycon to enable proper highlighting once rendered through MkDocs/Zensical?

Also the rendering issue in pyapp-kit/scenex#89 might simply be due to the lack of a blank line between the markup and the >>> block.

[gselzer]

Have you considered simply not indenting your >>> blocks

Yes, that's what pyapp-kit/scenex#93 does, however it is clear to me now that I am prone to doing this, and in the case others are as well I figured a fix upstream couldn't hurt 😅

and use https://github.com/pawamoy/markdown-pycon to enable proper highlighting once rendered through MkDocs/Zensical?

Umm, maybe, how is that different from the examples within the docstrings are being rendered now without pycon?

Also the rendering issue in pyapp-kit/scenex#89 might simply be due to the lack of a blank line between the markup and the >>> block.

Based on a quick change and a local Zensical build that doesn't seem to be part of the issue 🙁

[pawamoy]

Ah right the markdown-pycon extension makes it possible to write >>> anywhere, not just in Examples sections understood by Griffe.

Since de-denting such blocks in Examples sections works as expected, I'm gonna close this PR. I'll reconsider if more people run into this 🙂

Thanks again for your contribution!