fix(labrinth): reduce rows returned by project query

[fetchfern]

• This query is used to get all aggregated version fields for a project.
• The current query's DISTINCT doesn't actually deduplicate anything. It returns number of projects * versions per project * fields per version rows.
• All these rows are never deduplicated, even in code. aggregate_version_fields in ProjectQueryResult ends up storing all fields for every version, which is then cached on Redis. For projects with 1000+ versions that's megabytes of data.
• Anywhere in Labrinth that fetches project info pays that cost.

The new query doesn't group by version ID, so duplicate rows are never fetched. The query plan is basically the same, only the amount of rows returned is reduced.

MIN(vf.version_id) isn't strictly necessary here but the type used (VersionField) actually needs a version ID field for other purposes. We could use a different type eventually but that's a bigger change. Loader fields will soon be removed anyways

[github-actions]

Pull request changelog

App

Added

Changed

Deprecated

Removed

Fixed

Security

Website

Added

Changed

Deprecated

Removed

Fixed

Security

Hosting

Added

Changed

Deprecated

Removed

Fixed

Security

[aecsocket]

@Mergifyio queue

[mergify]

Merge Queue Status

• ✅ Entered queue — 2026-10-09 08:03 UTC · Rule: default · triggered by @aecsocket with the @mergifyio queue command
• ✅ Checks passed · on draft merge queue: checking #7901 on main (82a7b56) #7908
• ✅ Merged — 2026-10-09 08:16 UTC · at d994f0ab3de8cb7cd340465c8f3ed4a7df38340d

This pull request spent 13 minutes 26 seconds in the queue, including 13 minutes running CI.

Required conditions to merge

• github-review-approved [🛡 GitHub branch protection]
  • fix(labrinth): reduce rows returned by project query #7901
• any of [🛡 GitHub branch protection]:
  • check-success = Lint and Test
  • check-neutral = Lint and Test
  • check-skipped = Lint and Test