Add displaCy data structures to docs

[thomashacker]

Description

This PR adds a new section to the api/top-level#functions documentation for displaCy: Visualizer data structures. The section shows examples of the three data structures (Dependency Parsing, Entity Recognition, and Span Classification) and their types formatted as tables. To reduce cluttering, the nested dictionary types are formatted as a collapsable table.

The idea for this PR initially came from this discussion #10950
The types are based on this gist by @pmbaumgartner

Types of change

Documentation

Checklist

• I confirm that I have the right to submit this contribution under the project's MIT license.
• I ran the tests, and all new and existing tests passed.
• My changes don't require a change to the documentation, or if they do, I've added all required information.

[pmbaumgartner]

My in-line code comments are a little out of order so let me summarize them in a comment:

• I don't think we should link back to the Usage docs since we provide those examples in the sidebar
• Can we find a better way to make it clear the accordions have information about the sub-structures in each data structure?
• We should be consistent with the descriptions for the values/attributes where they exist in the spaCy API.

[thomashacker]

@pmbaumgartner What was the reason for removing the docs label from this PR? 😢

[pmbaumgartner]

@pmbaumgartner What was the reason for removing the docs label from this PR? 😢

I must have misclicked or used a keyboard shortcut accidentally. Sorry!

[svlandeg]

@pmbaumgartner: is your "changes requested" review still valid? (as discussed internally we don't usually use this feature as typically we'd go by the open review comments to determine whether something requires changes or not, but because you used it earlier I want to double check with you and not just dismiss the review)

Changes addressed / wrong review type

[pmbaumgartner]

@pmbaumgartner: is your "changes requested" review still valid? (as discussed internally we don't usually use this feature as typically we'd go by the open review comments to determine whether something requires changes or not, but because you used it earlier I want to double check with you and not just dismiss the review)

I wasn't aware this was still blocking. I dismissed the review, but haven't had time to review the recent changes yet, so I'll look again today.

[pmbaumgartner]

In the ENT example data structure, title has a value of None. Shouldn't its actual type be Optional[str] then? Same for Spans?

And it looks like title isn't a field on the deps data, only ents and spans, is that right? I think we might want to open an issue for that, if that's the case.

[thomashacker]

• Adjusted the type of the title key in the docs
• Added tests that check displaCy.render with manual data
• Implemented title key to dependency visualizations and adjusted docs

[pmbaumgartner]

👌 love the tests. I don't have feedback on anything else at this time.

[svlandeg]

Closing in favour of #12875 (PR redone to target master)