First steps: Joanna to test up-to-date dataset with MarcEdit, and decide if filtering non-top-level records is (a) sufficient and (b) efficient. If so, can close ticket.
If not, pass on to Simona with request to remove analytic records from the canonical dataset, to speed up this work.
Hand-editing of top-level records may continue (Joanna or library team), but some discussion of deduplication and merge process may still be required.