Choose the data model and write the GraphQL SDL

[erikhfj]

The schema is the contract between the client and the backend. Define it early so the group agrees on the data model before M2 implementation.

• Define the core catalog: album, artist, track and genre, including their relationships
• Define user-generated data: review and crate item
• Define searchAlbums with a cursor-based connection, total count and planned facet counts
• Define mutations for upsertReview, deleteReview and toggleCrate

Acceptance:

• schema.graphql describes the GraphQL contract for the catalog, search, reviews and crate.
• docs/data-model.md explains the important relationships and modelling choices.
• This PR deliberately does not implement PostgreSQL, a GraphQL server or resolvers. Those belong to M2 in PostgreSQL schema + migrations (FTS + triggers) #5 and GraphQL Yoga server with searchAlbums #6.

[erikhfj]

Before you open the PR: implement from the requirement, not by copying from archive/scaffold-prototype. The prototype is a reference you can check against when you get stuck. It is not a source to paste from.

The PR description must say what was done and what was tested. That is what we read when we review.