This is an interesting technical approach and fine for local usage but it does not work for Jupyter notebooks in general.

At its heart Jupyter is a protocol for messages sent between a front end (web browser) and a backend (Jupyter server) that may be running on the same or different machine, and can spawn new kernel processes (e.g. python) on the same or different machines. The communication here to update the plot is using direct communication between the front end and webagg's tornado implementation, bypassing Jupyter messaging. It works fine in a local dev setup where you have the Jupyter server and kernel running on the same machine and you aren't too bothered about security. But there are many Jupyter installations (e.g. JupyterHub) in which one or both of those are no longer true.

So probably it is best not to claim that works in Jupyter. At best you will end up having to deal with questions about tornado routing and security setup, and I would not wish that on anybody!

Note that to really get the webagg backend working in Jupyter one would have to extend it to use Jupyter messaging. This is precisely what the ipympl backend is, so that problem has already been solved.

I think your ultimate motivation here is to support blitting and avoid event accumulation in Matplotlib plots in Jupyter, in which case perhaps the best approach would be to continue with #27160. It looks like that was progressing well but stalled because you didn't have time to address the review comments. If you could get that working to your satisfaction here in the webagg backend that would be great. You can ignore ipympl, someone else (possibly me) will ensure the changes are incorporated there eventually.

Thank you very much @ianthomas23 for your detailed comment! I fully agree with your view on this.

Indeed ipympl is the proper way to implement a matplotlib backend for jupyter notebooks,
and finishing #27160 such that it works for ipympl as well is the way to go here.

I can try to find the time to finish #27160 properly, but to be honest, one reason why I've lost a bit of pace is because I've been
waiting since 18 October 2023 to get some help implementing the required ack message so that the solution can be used in ipympl (#19116 (comment)) ... which was my goal here from the start.

I can fix the python-side, but I don't have the time to dig into the javascript code since it's not my expertise...
If someone (maybe you?) could create a branch in ipympl that includes the necessary changes on the javascript side, I'd be happy to make sure that the webagg event accumulation fix is implemented such that it does it's job equally well for ipympl.

To be clear, an ack message like this one should be all I need: