While plotly/other libs are great for many tasks, they lack certain matplotlib features and performance.
mpld3 is a popular option, but interactions don’t work for many users (see here for example).
How do jupyter notebooks do this so seamlessly? Is there an avenue there worth exploring?