Add Real-World Project: Semantic Search Over Your Browser Bookmarks

[abderrahim-lectures]

Adds a new Real-World Projects lesson: Build Semantic Search Over Your Browser Bookmarks (backlog idea #17).

What's included

• docs/projects/bookmarks-semantic-search/index.md — full lesson in the house skeleton. The twist vs. the RAG projects: instead of chunking documents, each bookmark (title + URL + folder path) is already the atomic record, so the challenge shifts to parsing a real export format (standard library html.parser, no dependencies) and to a keyword-vs-semantic search comparison.
• docs/projects/bookmarks-semantic-search/_category_.json
• examples/bookmarks-semantic-search/ — runnable, no-API-key tool: parse_bookmarks.py, build_index.py, search.py, compare.py, plus bundled sample_bookmarks.html, notebook.ipynb, README, pyproject/uv.lock.
• Registration: src/data/projects.ts, docs/projects/index.mdx, src/pages/index.tsx (English only — i18n in a follow-up PR).

Verification

• Pipeline tested end-to-end locally: parsed the bundled sample (30 bookmarks, correct nested folder paths), embedded, and semantic search ranks train_test_split docs top (0.895) for a query with zero shared title words; keyword search finds it only via exact title hits.
• npm run typecheck and npm run build pass (all 4 locales, no broken links).

Closes #272