Description
Hello Dev!
I have been using Plotly's Python library and encountered an issue when trying to set the figure config while preserving on_change callbacks on the same figure object.
The root cause is that _config is currently a private attribute in the BaseFigure class, and the only api way to modify it is through the fig.show(config=config) method call. However, show() rerenders the figure and won't carry existing on_change listeners.
Therefore, in theory, using __setattr__ to modify the _config attribute directly instead of using .show(config=config) would make both setting on_change() and setting config (e.g. scrollZoom = True) work well at the same fig, and this has been proven to be the case in testing.
So I suggest to make _config a public attribute that users can directly set it without calling show() or using setattr.
Many thanks for your supper awesome work on Plotly!