feat: Introduce component ids

[bprusinowski]

Closes #1696

This PR introduces a new, component id concept to Visualize, which replaces the current component iri convention.

Component id is made by combining unversioned cube iri with unversioned component iri, which is a unique combination even in case we merged two cubes with the same component iris. This means that we can handle these cases correctly, while leaving our business logic mostly intact (and in some cases, even simpler, as we don't need to filter by cube iri and component iri anymore, but instead simply use component id).

This combination of iris is done on the server side and sent to the client, which doesn't need to know about this fact, but simply keep the current logic (chart config migrations are the only exception, as we need to introduce unversioned cube iris to relevant parts of the config dynamically to make sure old charts work).

By proxy, it was necessary to make the migrations async, so we can fetch unversioned cube iri when needed – as immer doesn't support / recommend asynchronous operations, I removed it.

ℹ️ Dimension id docs

🔎 Visual regression testing docs

How to test

🚨 Preferably, the implementation should be tested holistically, as essentially every part of Visualize was affected in one or another way.

1. Go to this link.
2. Click on Add dataset.
3. Add Market figures milk and dairy products - Monthly producer prices dataset.
4. Open Vertical axis.
5. ✅ See that it's possible to toggle between two Price dimensions that have the same iris and names, but come from different cubes.
6. Change chart type to Table.
7. ✅ See that there are duplicated dimensions visible in the Table (that have the same iris and names, but come from different cubes).

How to reproduce

Repeat the above steps on PROD environment and see that ✅ points are broken ❌.

---

• Add docs to feature pages in Notion (both for componentId and visual regression testing)
• Add tests
• Check E2E tests locally
• Check if all old charts work (checked local /_charts page and cross-compared 500 PROD charts on main vs this branch)

[vercel]

The latest updates on your projects. Learn more about Vercel for Git ↗︎

Name	Status	Preview	Comments	Updated (UTC)
visualization-tool	✅ Ready (Inspect)	Visit Preview	💬 Add feedback	Nov 20, 2024 2:50pm

[ptbrowne]

Thanks a lot Bartosz for this massive effort 💪

I see no blockers here, I have mostly put nits and comments.

• The use of branded types would make sure we do not pass a dimension iri where we expect a dimension id.
• I think it would be a good idea to perform screenshot testing against production charts to make sure they are ok (load the production charts into local database, screenshot the chart on main, switch to this branch rescreenshot, check if they are the same)