feat: use HTML parser (scraper crate) instead of regex for URL extraction

[ChefControl]

Summary

• Replace custom regex (r"https?://[\w\-.]+(?::\d+)?") with the scraper crate to parse HTML and extract URLs from <a href="..."> tags
• Resolve relative URLs against the page's base URL using the url crate
• Remove regex dependency from workspace and shared crate, add scraper and url
• Update callers in feeder and manager to pass base URL to the new extract_urls(html, base_url) signature
• Expand test suite from 6 to 14 tests covering relative paths, query strings, fragments, protocol-relative URLs, non-http scheme filtering, and invalid base URL handling

Test plan

• All 40 shared crate tests pass (cargo test -p shared)
• Full project compiles (cargo check)
• Manual integration test: crawl a site and verify URLs with paths/query strings are now discovered

Closes #26

🤖 Generated with Claude Code