fix(zhihu/download): author extraction, zhida links, redirect detection

[yrom]

Summary

Fix multiple issues in the zhihu download adapter.

Fixes #40 — Author extraction returns "unknown"

Root cause: The combined CSS selector .AuthorInfo-name, .UserLink-link matches elements in DOM order. A .UserLink-link element appearing before .AuthorInfo-name in the DOM returns empty text, causing the fallback chain to skip the real author name.

Fix: Split into separate querySelector calls with proper fallback order:

1. .AuthorInfo-name (primary)
2. .UserLink-link (secondary)
3. meta[itemprop="author"] (stable meta tag fallback)
4. meta[name="author"]
5. js-initialData JSON parsing (last resort)

Fixes #42 — Template variable not filled + wrong article

Root cause 1: On failure (content element not found), the evaluate step returned an array [{...}] instead of a plain object {...}. The download step's ${{ data.title }} template couldn't resolve, leaving the literal string in the output.

Fix: Return a plain object with all required fields (filename, imageUrls, content, output) on error, matching the success case structure.

Root cause 2: When an article URL is removed/not found, Zhihu redirects to the homepage. The adapter would then scrape the homepage content instead of reporting an error.

Fix: Detect redirect by checking location.hostname and location.pathname after navigation. Return a clear "Article not found" error.

Additional improvements

• Strip zhida.zhihu.com links: These are Zhihu's internal AI search links that add noise to the markdown output. Now only the display text is kept.
• Increase settleMs from 3000 to 5000 for more reliable page loading.
• Add path column to output table so users can see the downloaded file path.

Test plan

• autocli zhihu download "https://www.zhihu.com/question/351504112/answer/2027391723035275294" — author correctly shows "NGINX洪志道" (was "unknown")
• autocli zhihu download "https://zhuanlan.zhihu.com/p/60954299" — shows "Article not found" error (was ${{ data.title }} literal)
• autocli zhihu download "https://zhuanlan.zhihu.com/p/61154299" — correct article fetched
• autocli zhihu download "https://www.zhihu.com/question/2040832649963508039/answer/2041160812719498582" — new URL format works