Add dataset-related filters to chart editor variable selector

[mlbrgl]

Summary

Add additional dataset-related filter dropdowns to the chart editor's variable selector, similar to the existing "Namespaces" filter.

Context

As part of the search-powered data collections work (PR #5939), we're exposing four dataset-related attributes on chart records in Algolia:

• namespace
• name (dataset name)
• version
• producer

These same attributes could be used as filters in the chart editor variable selector to improve the indicator selection workflow.

Current State

The variable selector (adminSiteClient/VariableSelector.tsx) currently has:

• A free-text search field
• A "Namespaces" multi-select dropdown filter

Requested Enhancement

Add additional filter dropdowns for:

• Dataset product - filter variables by dataset name
• Dataset version - filter variables by version
• Producer - filter variables by producer

Reference

• Related PR: feat: add dataset dimensions to chart records #5939

[ikesau]

I think we could be more ambitious here

@danyx23 could you chat with the D&R team to figure out how they'd like this modal to be better?

[danyx23]

Pain points described in the D&R pain points session:

• Filtering by dataset would be useful - could be either the ETL path or a more fine-grained. ETL Path is maybe easiest?
• Scrolling is weird and feels broken/janky which makes you miss thigs
• Cases when a dataset has a lot of indicators (e.g. GBD), it would be great if we could use the dimensions information and render grouped checkboxes based on those. We have this information in the DB, once you are filtered down to one dataset this could be really useful
• It might be useful to switch display between titles and full etl paths
• Add a filter that hides all indicators that are not used at all in published charts so far
• Might be useful to be able to filter by dataset id in addition to etl path
• Selection settings should persist across open/close (but maybe add a clear button then) because you sometimes need to add a lot of very related indicators