migrations: Revert backfilling repository_id

In [1] a migration was added to backfill the repository_id on the storage_repositories table. That backfill has been using a single UPDATE query to update all records at once. But because there is a TRIGGER on that table it caused a PG_NOTIFY to be sent with a payload that exceeds the maximum allowed by PostgreSQL, and the migration to fail on large datasets.

That commit is part of [2]. After that commit, the queries can be modified to JOIN tables on the repository_id, instead of on the virtual_storage + relative_path. But those changes are not merged yet. Therefore this change removes the migration to unblock installations with large datasets, until a proper solution to backfill the records is found.

1. 2bbec66c (Link existing database record via repository ID, 2021-09-01)
2. #3485 (closed)