v4.2.0

Changed

• scrape_as_markdown and scrape_batch use Bright Data's Fetch API
  instead of Web Unlocker. There is no JavaScript rendering.
• Both tools return JSON: url, title, format, content, plus
  final_url after a redirect and a note when relevant.
  scrape_batch returns an array in input order; a failed URL is
  {url, error}.
• New format param: markdown (default), text, html or
  ld_json. If the page has no data in that format, markdown is
  returned with a note.
• scrape_as_markdown stops at 45 s with a timeout error.
  scrape_batch sends all URLs in one request, cancels at 35 s and
  returns the pages that finished; the rest get a timeout error.
• Clear one-line errors: page_not_found, fetch_failed,
  invalid_url, rate_limited, empty_page, unsupported_content
  (PDF/binary), no_data, timeout, auth_failed, busy (429, never
  retried), service_error (one retry on 5xx), network_error.
• Page content is cut at 200,000 characters (scrape_as_markdown) or
  100,000 per page (scrape_batch), with a note.
• Markdown is no longer stripped of headings.

Removed

• remark and strip-markdown dependencies.