Add Startpage backend and update scraper HTTP fingerprint

[DavidTeju]

Summary

• Add Startpage as a new search backend — Startpage proxies Google results as static HTML, making it scrapeable without JavaScript. Google and DuckDuckGo now serve JS-only challenge pages that no static HTTP client can bypass.
• Make Startpage the default for new installs (existing user configs are preserved)
• Update scraper HTTP fingerprint — replace the hardcoded 2012 Firefox 11 User-Agent with a rotating pool of 8 modern browser UAs and add browser-realistic headers (Accept, Accept-Language, Upgrade-Insecure-Requests, Sec-Fetch-*)

Context

Both Google and DuckDuckGo have escalated their bot detection beyond User-Agent checking:

• Google now returns a JS shell page with <noscript> redirect to /httpservice/retry/enablejs — no search results in the HTML
• DuckDuckGo returns an anomaly.js challenge form that requires JavaScript to solve

These changes affect all HTTP clients equally — neither the old Firefox 11 UA nor modern browser UAs receive actual search results. The scraper backends need a fundamentally different approach.

Startpage (startpage.com) is a privacy-focused search engine that proxies Google results and serves them as plain, scrapeable HTML. It:

• Uses the same (site:X OR site:Y) query format already used by the Google/DDG scrapers
• Returns direct href links to StackOverflow questions (no redirect wrappers)
• Works reliably from reqwest with browser-like headers
• Returns 10 results per page with the a.result-title CSS selector

Changes

Commit 1: Update scraper HTTP fingerprint

File	Change
src/stackexchange/mod.rs	Add SCRAPER_USER_AGENTS pool (8 modern UAs), select_scraper_user_agent(), scraper_headers() with browser-matched headers, scraper_client() builder, 6 unit tests
src/stackexchange/search.rs	Use super::scraper_client() in search_by_scraper() instead of bare Client::new() with single UA header

Commit 2: Add Startpage backend

File	Change
src/stackexchange/scraper.rs	Add Startpage struct implementing Scraper trait with a.result-title selector
src/config.rs	Add Startpage variant to SearchEngine, make it #[default]
src/cli.rs	Add "startpage" to search engine value parser
src/stackexchange/search.rs	Add SearchEngine::Startpage match arm
test/startpage/exit-vim.html	Test fixture (real Startpage response)

Zero new dependencies. UA rotation uses std::process::id() modulo instead of rand.

Test plan

• cargo fmt --all --check passes
• cargo test — all 26 tests pass (8 new: 6 for UA/headers, 2 for Startpage URL + parser)
• Smoke test --lucky -e startpage "how to exit vim" — returns full SO answer
• Smoke test --lucky -e startpage "rust reverse string" — returns full SO answer
• Smoke test --lucky -e stackexchange "rust reverse string" — still works (regression)
• Existing Google/DDG backends unchanged and still available for users who want them

Closes #16. Addresses #32.