Refine feature selection scope

[amirrr]

Internal representation

[{'name': 'paper', 'children': ['title', 'experiments']}]
[{'name': "experiments", 'children': ['name', 'description', 'conditions']}]
[{'name': "conditions", 'children': ['name', 'description', 'type', 'message', 'behaviors']}]
[{'name': "behaviors", 'children': ['name', 'description', 'priority', 'focal']}]

GPT response:

{
  title: 'something',
  experiments: [
    {
      adults: '--',
      age_mean: '--',
      age_sd: '--',
      compensation: '--',
      conditions: [
        {
          description: '--',
          message: '--',
          name: '--',
          type: '--',
          behaviors: [
            {
              description: '--',
              focal: '--',
              name: '--',
              priority: '--',
            },
          ],
        },
      ],
      description: '--',
      name: '--',
    },
  ],
}

[markwhiting]

Assuming a flat array of features that include at least name and parent:

1. get the parents and their order (parent's parent > parent) so we can render the buttons for selecting which parents to summarize.
2. Expand all features that are children of a selected parent.
3. Summarize all features that are children of a unselected parents by showing the count of unique rows if their parent was expanded.

This returns:

1. An array of parents to be used as buttons
2. An array of features ordered by parents with a new property summarized.

Then, compute how each row should appear in the table based on that.

1. Parents above expanded children should be summarized by showing the name feature of the parent level, e.g., conditions.map(condition => condition.name).
2. Parents below the last expanded parent should appear as a number showing how many rows would appear if that parent were expanded, e.g. behaviors.length