PR summary

Covering the actual use case for MNE-Python I was investigating when I hit #32264 : needing to update contours inside an animation. This PR adds ContourSet.set_data so that contours can be blitted properly rather than removed and readded.

This is really "issue as PR" -- I wasn't sure how many lines of code would be needed to make it work, so I figured I'd just check by implementing the changes... So I'm happy to close if this isn't wanted or refactor heavily if there is a better way, etc. It didn't end up being so bad since _process_args could be reused -- most of the new lines are unit tests actually. Related issues:

• Animation of contourf becomes extremely slow #6985 (noted the limitation, as did this SO post and this Discourse post)
• ENH: Collection.set_paths #26342 (precedent/similar)

One intentional omission is support for setting levels. I think that should be a follow-up PR if desired -- it interacts with colorbars etc. which is a harder problem. (And maybe should be solved more generally with a .set_levels instead!) Also fixes a small TriContour bug I encountered when adding a **kwargs check (see lib/matplotlib/tri/_tricontour.py). I also added a helper to reset the object state in case the set_data fails after some attributes are already set (which can happen inside _process_args), put in cbook so that other Artists could potentially use it down the line.

AI Disclosure

I had Claude Opus 5 draft the code changes. I iterated through them until it looked like what I would have produced on my own (with better/more complete testing actually).

PR quality check

• Use an expressive title, e.g. "Fix title font property precedence"
• New and changed code is tested
• Plotting related features are demonstrated in an example
• New features and API changes have release notes
• Documentation complies with general and docstring guidelines