v1.4.0

This release replaces the serial two-phase crawl-then-scrape design with a single
merged task group. Crawling and scraping now run concurrently: each page is
scraped the instant it finishes crawling, without waiting for the rest of the
site.

What changed

• crawlAndScrape is now a single pipeline. Previously the method crawled
  every page first, collecting all HTML bodies into an array, and only then
  started scraping. Now crawl and scrape tasks share one ThrowingTaskGroup —
  as soon as a page is crawled, it's immediately enqueued for parsing, metadata
  extraction, and markdown conversion.
• Lower memory. HTML bodies are held only until their scrape completes,
  rather than accumulating every page before any scraping begins.
• Sooner first result. The first ScrapedPage is yielded as soon as the
  first page finishes crawling AND scraping, not after the entire crawl finishes.
• Bounded scrape concurrency. The number of concurrent scrape tasks is
  capped at CrawlConfiguration.concurrency (default 8), preventing CPU-bound
  parse work from piling up when crawling outpaces scraping.