Default db_path to data/<dataset-id>/netflow.sqlite, the location the
dashboard auto-discovers. When default_start_date is unset, store the
earliest ingested local day after each pipeline run instead of a
hardcoded date. Explicit configuration still wins for both fields.