[ENH] eCharts as alternative plotting backend

[borolepratik]

Proposed new feature or change:

Would be nice to have echarts as alternative backend for interactive visualisations.

https://github.com/pyecharts/pyecharts

[Casper-Guo]

Todos are similar to #837

[theOehrly]

I think that given the popularity of the (py)echarts, this is a valid request too. But also, I want to be careful with adding more and more plotting "backends" for two reasons. First, I want to make sure to keep the API clean. This should not generate a mess in the plotting submodule. And second, all of this adds some degree of maintenance effort.

For example, there might be more candidates as well, for example:

• https://github.com/vega/altair
• https://github.com/has2k1/plotnine
• https://github.com/bokeh/bokeh

Adding just about everything, is probably not the solution. So maybe its necessary what package provides a clear advantage and what package is very similar to another one. Or maybe some of these simply don't play well with the requirements that we have in FastF1.

In any case, I'd probably like to wait for @Casper-Guo to create a PR for #837. That might give a better base for discussing the viability of multiple plotting backends with regard to implementing this cleanly.
And I most probably won't be working on that. So someone else will need to do the implementation.

[Casper-Guo]

Sorry about the delay on this.

First, I want to make sure to keep the API clean. This should not generate a mess in the plotting submodule. And second, all of this adds some degree of maintenance effort.

My proposed alternative would be to do these things as tutorials instead. Supporting different backends is mostly hardcoding logic. Even supporting Matplotlib seems mostly a matter of legacy now. This will be beneficial from a maintenance burden perspective and also makes it easier for other contributors to introduce visualization libraries that they are familiar with.

[theOehrly]

Tutorials might also be a good way to add "support" for various plotting tools. I do like that idea.

Having one or maybe two directly supported backends is a legacy thing, yes. But also, it allows to easily add examples where boilerplate plotting code does not get in the way of the example itself. And it allows new users to quickly get started and prototype visualizations. Therefore, I'd say that plotting support as it is, will certainly be kept for the foreseeable future.