Add BBM to mark packages_helm_metadata_caches as pending_destruction

What does this MR do and why?

In the previous MR, we fixed the bug to update helm_metadata_cache for all channels when package gets deleted.

Because of the bug, we disabled the feature flag packages_helm_metadata_cache, to be able to do the rollout again, we need to clean up the packages_helm_metadata_caches table, otherwise, users who already had packages_helm_metadata_caches records for the given project and channel, will get the legacy/wrong metadata cache.

This MR introduces a post-deployment migration to mark all the records in packages_helm_metadata_cache as pending_destruction status, since the table has LFK on project_id, it will trigger the deletion of the record as well as the file in object storage.

Database

Currently, we have 4748 records in packages_helm_metadata_caches table.

Raw query

We mark packages_helm_metadata_caches in batch.

UPDATE
    "packages_helm_metadata_caches"
SET
    "status" = 2
WHERE
    "packages_helm_metadata_caches"."id" BETWEEN 11 AND 16
    AND "packages_helm_metadata_caches"."id" >= 11

Query plan: https://console.postgres.ai/gitlab/gitlab-production-main/sessions/43023/commands/131552

References

• Feature flag

Screenshots or screen recordings

N/A

How to set up and validate locally

1. Switch to this branch
2. Enter Rails console and create packages_helm_metadata_caches records

   sql = <<-SQL
       INSERT INTO packages_helm_metadata_caches 
           (project_id, channel, object_storage_key, size, file, status, created_at, updated_at)
       VALUES 
           (2, 'channel-0', 'key-0', 401, 'index.yaml', 0, NOW(), NOW()),
           (2, 'channel-1', 'key-1', 401, 'index.yaml', 0, NOW(), NOW()),
           (2, 'channel-2', 'key-2', 401, 'index.yaml', 0, NOW(), NOW()),
           (2, 'channel-3', 'key-3', 401, 'index.yaml', 0, NOW(), NOW()),
           (2, 'channel-4', 'key-4', 401, 'index.yaml', 0, NOW(), NOW()),
           (2, 'channel-5', 'key-5', 401, 'index.yaml', 0, NOW(), NOW()),
           (2, 'channel-6', 'key-6', 401, 'index.yaml', 0, NOW(), NOW()),
           (2, 'channel-7', 'key-7', 401, 'index.yaml', 0, NOW(), NOW()),
           (2, 'channel-8', 'key-8', 401, 'index.yaml', 0, NOW(), NOW()),
           (20, 'channel-9', 'key-9', 401, 'index.yaml', 0, NOW(), NOW());
   SQL
   
       ActiveRecord::Base.connection.execute(sql)

3. Check records, the statuses should all be default

   Packages::Helm::MetadataCache.all.map(&:status)

4. Exit Rails console and run migration

   rails db:migrate:up:main VERSION=20250829141659

5. Check background_migrations page (http://gdk.test:3000/admin/background_migrations)
6. If exists, enter Rails console to check the status again

   Packages::Helm::MetadataCache.all.map(&:status)

   All of them should become pending_destruction

MR acceptance checklist

Evaluate this MR against the MR acceptance checklist. It helps you analyze changes to reduce risks in quality, performance, reliability, security, and maintainability.

Related to #566574 (closed)