Get object pages

Retrieves the raw text of a processed vault object split by page. The object must have completed ingestion before pages can be retrieved — for PDFs this requires the OCR pipeline to have finished writing the per-page sidecar, so freshly uploaded PDFs return 400 with the current ingestionStatus until processing completes. For PDFs this returns the per-page OCR text. For plain text files (txt, md, source code, court reporter transcripts) the text is split using right-aligned page-number markers when present (preserving the original document numbering, including continuations like Volume 2 starting at page 234), falling back to form-feed (\f) page-break characters, and finally a single page if neither signal is present. Use the optional start and end query parameters to fetch a specific inclusive page range. Pages with no text are omitted.