Fix dropdown option spacing in markdown output

Summary

This PR fixes a bug where HTML dropdown (<select>/<option>) elements were being converted to markdown with no spacing between options, resulting in concatenated text like "Option1Option2Option3" instead of "Option1, Option2, Option3".

Changes:

• Added a preprocessSelectElements() function that runs before HTML-to-markdown conversion
• Select elements are replaced with comma-separated lists of their option values
• Filters out placeholder options (empty value="" or text starting with "Select")
• Skips disabled options
• Normalizes whitespace in option text
• Added 6 comprehensive unit tests covering various scenarios

The fix works uniformly across both the Go converter and TurndownService fallback by preprocessing the HTML before conversion.

Review & Testing Checklist for Human

Risk Level: Yellow (Medium risk - modifies core parsing logic with limited end-to-end testing)

• Test with the actual problematic URL - Scrape https://renovahospitals.com/ through the Firecrawl API and verify the specialty dropdown options are properly formatted with commas/spaces
• Check for performance impact - Test with large HTML documents (especially those with many or large select elements) to ensure the preprocessing doesn't significantly impact scraping speed
• Verify no regressions - Test scraping various websites that use different HTML structures to ensure the preprocessing doesn't break other conversions or cause unexpected side effects
• Test edge cases - Try dropdowns with: (1) options containing commas in their text, (2) very long option lists (100+ items), (3) nested or multiple selects on the same page

Test Plan

# Test the problematic URL
curl -X POST https://api.firecrawl.dev/v1/scrape \
  -H "Authorization: Bearer YOUR_API_KEY" \
  -d '{"url": "https://renovahospitals.com/", "formats": ["markdown"]}' \
  | jq '.data.markdown' | grep -A 5 "Speciality"

Expected: Should see "Anaesthesiology & Pain Management, Arthroplasty And Sports Medicine, Cardiology, ..." with proper comma-space separation.

Notes

• All 11 unit tests pass locally (including 6 new tests for select/option handling)
• Fixed an unrelated test expectation that was using the wrong markdown list format
• The placeholder detection regex /^select\b/i might need refinement based on real-world usage patterns
• Performance impact should be minimal since cheerio is already a dependency and the preprocessing is lightweight, but worth monitoring

Session: https://app.devin.ai/sessions/b5d46cd2ec9d4285ad2cf6c9d1bc2027
Requested by: Micah Stairs (@micahstairs)

Summary by cubic

Fixes markdown conversion for HTML dropdowns by preprocessing select/option elements and outputting comma-separated lists. Prevents concatenated text like "Option1Option2Option3" across both the Go converter and Turndown fallback.

• Bug Fixes
  • Replace select elements with comma-separated option text.
  • Ignore placeholder (empty value or text starting with “Select”) and disabled options; normalize whitespace.
  • Add 6 unit tests for select/option handling; adjust one list test expectation.

^{Written for commit 6daee32. Summary will update automatically on new commits.}