Update MigrateRecordsToGhostUserInBatchesService retry logic to prevent from being blocked at one user

As an additional update from the issue, we can update the retry logic for MigrateRecordsToGhostUserInBatchesService by story a retry attribute for each row/user to avoid getting blocked if that deletion fails for a reasons. @ahegyi recommended:

We can use the same pattern that we have for loose foreign keys.

1. Add the consume_after column to the ghost_user_migrations table.
2. Set the consume_after value to created_at by default (could be NOW() as well) when the ghost_user_migrations record is created.
3. When a user deletion fails, catch the timeout error and update the consume_after with Time.current + 30 minutes.

Additionally, we should retrieve a batch by ordering the rows by consume_after ASC