Create an asynchronous job over a list (datavask, reverse, bbr, dagi, miljoe)

The list arrives as JSON { kind, rows, columns?, srid?, label?, start? }, a multipart file (CSV, NDJSON or a JSON array) with the same fields as form fields, or a raw text/csv / application/x-ndjson body with them as query parameters. columns maps roles (adresse, vej, husnr, postnr, by, x, y, bfe, adgangsadresse_id) to your column names; left out, they are detected from the names. Rows are worked in the background in modest chunks. Every uploaded row is counted against the quota at upload and handed back if it does not succeed: ONLY successful rows are billed. start: false makes a draft to fill with POST /v1/jobs/{id}/rows and start with POST /v1/jobs/{id}/start. A job.completed webhook fires when the last row is done. Rows per job follow the plan: 250 on Free, 10000 on Starter, 100000 on Pro, 500000 on Scale, 1000000 on Enterprise (402 over). One upload is at most 10 MB; upload a longer list in parts.