Download the cost report as a CSV export

Streams a CSV export of the referenced Cost Report — the synchronous CSV twin of the GET .../cost-reports read: one row per time bucket, one column per resolved dimension, built from the same groupBy/period/granularity query dimensions as the report itself. The download does not mutate any resource. Validation: scopeId must be accessible and costReportId must reference a cost report within that scope (else 404 NOT_FOUND); groupBy, period and granularity must be valid enum members when supplied (defaults: service/30d/daily) — otherwise 422 VALIDATION_FAILED; the caller must hold a billing-capable role (owner|admin|contributor|reader|billing), else 403. Async & state: Synchronous read; no state change. The CSV is rebuilt from the live cost dataset and streamed inline; no Operation is created and no resource.updated SSE frames are emitted. Errors: 401 unauthenticated; 403 insufficient role; 404 unknown scope or cost report; 422 malformed query dimensions; 500 internal failure. Idempotency/Concurrency: Safe, cacheable; no idempotency key required. Repeated downloads with the same dimensions return byte-identical CSVs for an unchanged dataset.