Submit a CSV of liens to import from the public record (one lien per row). Always processed asynchronously (202 + batch id); poll GET /lien_submissions/import/batches/{batch_id} and download per-row results via /lien_submissions/import/batches/{batch_id}/download. Batches are capped at 40,000 rows. Columns: state, filing_number, filing_name (required: state, filing_number). Only the file as a whole is validated on upload — headers, encoding, size and row count; a row the parser cannot map is reported per-row in the results instead. Rows are identified by their ORIGINAL line number in the uploaded file (the header is line 1, and skipped blank lines still take up their line). One row per line: a quoted field may not contain a line break, and a file with one is rejected naming the line the row starts on.