fix: drop markdown_v2 access in BasicWebScraper.extract

[obchain]

What

Apply the same markdown_v2 → markdown cleanup that #52 lands in WebScraper.extract to BasicWebScraper.extract. The two paths had byte-identical broken accesses; this PR brings the second one in line.

Why

src/opendeepsearch/context_scraping/basic_web_scraper.py:50-51:

if result.success:
    extraction_result.raw_markdown_length = len(result.markdown_v2.raw_markdown)
    extraction_result.citations_markdown_length = len(result.markdown_v2.markdown_with_citations)

crawl4ai 0.5.x removed markdown_v2 and replaced its __getattr__ with a hard AttributeError (full trace already in #34). BasicWebScraper is currently dead — crawl4ai_scraper.py only imports the ExtractionConfig dataclass that lives in the same module — but the class is public and the bug is real for anyone reaching for it.

Closes #59

How

src/opendeepsearch/context_scraping/basic_web_scraper.py:

• compute markdown_obj = getattr(result, 'markdown', None) or getattr(result, 'markdown_v2', None) once per result
• gate the length bookkeeping on result.success and markdown_obj is not None
• read raw_markdown / markdown_with_citations with getattr(..., '') or '' so a partially-populated MarkdownGenerationResult doesn't crash either

Kept narrowly scoped. Did not delete the unused BasicWebScraper class itself — that's a separate cleanup the maintainers may want to weigh.

Testing

• python3 -m py_compile src/opendeepsearch/context_scraping/basic_web_scraper.py — clean
• grep -n markdown_v2 src/opendeepsearch/context_scraping/basic_web_scraper.py — only the intentional fallback line remains
• Mirrors the manual repro path documented in fix: use crawl4ai result.markdown instead of removed markdown_v2 #52 (with crawl4ai 0.5.x and a fake result whose markdown_v2 accessor raises, the unpatched code re-raises; the patched code now reads result.markdown.raw_markdown).