[DOCS] Mega Thread for Top Doc Improvements by 1.0

[henrydingliu]

We've come along way on the doc in the past couple of months. I'd like to focus our efforts in the coming months to make our doc site 1.0-worthy. Let's use this thread to align on what main doc-related initiatives we should accomplish by 1.0. I'm going to give this "Poll question" template a try. Here are my top 3 for 1.0. I will elaborate on each in a comment. I'd love to hear from everyone.

Out of scope for this thread (has separate threads):

• Friedland recreation
• doctest warnings

[henrydingliu]

Robust testing pipeline (notebook fails, etc.)

A lot of us have been operating under the assumption that failing notebook cells would trigger a unittest fail. That is not the case today. Should we actually implement it? Or should we actually let doc build fail? Either way, there's still nuances like rtd vs unittest dependencies that introduce hidden (or extra) fails.

[henrydingliu]

Docstrings for all Triangle methods

With examples built into docstrings for all classes, the next frontier is all the methods under Triangle (and its various mixins).

[henrydingliu]

Cleaner TOC

the API TOC is getting pretty cluttered. how can we clean it up? and with the triangle page displaying so many methods now, should we make each method into a page?

[genedan]

For Triangle, there should be one method per page. Inherited methods like TriangleSlicer.iat[] should appear as Triangle.at[], since that's how the user sees it. I think for estimators it's ok to have multiple methods on a page since there tend to not be a lot of them.