feat: Deep Research crawl + selective source import & HTML preview

[PengyiZhang]

Summary

• Auto-crawl full page content for all Deep Research search results in parallel with LLM synthesis
• Allow users to selectively import crawled pages as source files into raw/sources/ for standard LLM ingest
• Add HTML file preview via sandboxed iframe for .html source files

Changes

• web-crawler.ts: URL crawler with concurrency=4, 15s timeout, regex-based article/main/body extraction
• research-store: Add crawledPages, crawlProgress, selectedUrls state and actions
• deep-research.ts: Auto-crawl after search + importSelectedSources() for selective import
• research-panel.tsx: Add checkboxes, crawl progress bar, select all/import button to Sources section
• file-preview.tsx: Add HtmlPreview component using sandboxed iframe
• file-types.ts: Add html file category
• i18n: Add crawl/import translations in en.json and zh.json

Test plan

• Start a Deep Research task, verify crawl runs automatically after search with progress indicator
• After crawl completes, select sources and click Import Selected, verify files are written to raw/sources/ and enqueued for ingest
• Click an imported .html source file, verify rendered preview instead of raw code

🤖 Generated with Claude Code