v4.8.0

Added

• scrape_as_markdown can read a long page in parts. A cut result has a
  next field with snapshot_id and start; calling
  scrape_as_markdown again with them returns the following part from
  the saved snapshot, without fetching the page again and at no extra
  cost.
• The note on a cut page says how much was returned and how big the
  whole page is, e.g. Cut at 39,996 of 472,270 characters (8%).

Changed

• scrape_batch caps the whole result at 40,000 characters (was
  40,000 per page), shared between pages: short pages are returned
  whole, longer ones split what is left. Each cut page has its own
  next for scrape_as_markdown.
• Pages are cut at the last line break before the limit, so a link or
  table row is not split.