Create model-dataset matching functionality in sci.vision

[ots22]

• clickable links to models/datasets in the table
  • could start with https://react-data-table-component.netlify.app/?path=/docs/expandable-basic--basic
• minimal 'model card' (showing description, installation, usage of the model) (e.g. as in Proposed scivision workflow for UI users with minimal coding exp. #325)

Issues/PRs to look at/finish before this one

• Gather requirements for a data loader for scivision (potential replacement for Intake) #480
• Refactor front end javascript #513
• [ENH] Add flag/tag "Installable" and/or "loadable" for models/datasets that work with scivision python package #428

PR for the below

• [Experiment] Python code display in front end #514

TODO 27th Jan '23:

• Update the model pages to include model cards (similar to huggingface) that shows both how to install and how to run with scivision
  • Install could be based on the contents on an install.md in the model repo (maybe .scivison/install.md)
  • One difficulty with this approach is that people with non-write access to a scivision model GitHub repo, you can't get a raw link for the markdown. Could we give scivision maintainers write access for all model repos? Would this even work with the production version?
  • Generate Python commands needed for chosen model or dataset
• Implement the matching idea as a new UI view as in the screenshots below
• Filter such that users only see datasets compatible with a model and vice versa
• Also enable filtering on a task (or domain, or other tag) in both table/grid view and in the new matching view to narrow the models (or datasets) in the page

[edwardchalstrey1]

[edwardchalstrey1]

As a first step of this, I think it would be good to improve the model/dataset tables themselves:

• Thumbnails for datasources: Include thumbnail images in datasource and model catalog entries #204
• Clickable links to GitHub repos
• Expandable descriptions

[ots22]

PR inbound with some improvements!

[edwardchalstrey1]

PR inbound with some improvements!

Ok nice, I'll take a look in a bit! Images on slack look great

[ots22]

See #383