Conda recipe: should we install any front-ends automatically?

Given that Altair now works on several different front-ends, we need to make some choices for the conda forge recipe.

In particular, when someone types conda install altair -c conda-forge, should it also install jupyter notebook and/or jupyterlab?

On one hand, it would be nice to have a single install command that works for new users, and we could direct people who don’t want one of these renderers to be installed to use pip instead.

On the other hand, it might be annoying to people who might not expect optional dependencies to be installed and/or updated when they conda install altair.

Any thoughts?

[xhochy]

Using Altair outside of the Notebook environment as a plotting backend for web apps is my main use case. In that environment it is annoying to depend on jupyter as it pulls in a huge set of requirements. The more lightweight the default dependencies are, the better for my specific use case. I can understand the current situation to make it simple for new users but it might be good to keep in mind that not all Altair users will be using it using inside notebook.

[jakevdp]

In the end, we chose not to depend on any frontends, which I think was the right choice.