fix(web-search): accept <div> result titles in Brave keyless scraper

[tongshu2023]

Summary

Fixes #687 — the keyless Brave web-search path returns 0 results because Brave moved the result title from a <span> to a <div>.

Problem

parseBraveSearchHtml matched the title with a <span class="search-snippet-title">-only regex. Brave's current markup is:

<div class="title search-snippet-title line-clamp-1 …" title="Title">Title</div>

So titleMatch was always null, if (!title) continue; skipped every snippet, and the scraper returned an empty list even though the page contained 20 data-type="web" blocks (see the live evidence in #687).

Fix

Accept either tag for the title element, using a backreference so the open/close tags stay consistent:

/<(span|div)[^>]*class="[^"]*search-snippet-title[^"]*"[^>]*>([\s\S]*?)<\/\1>/i

Tests

• Updated the parseBraveSearchHtml and searchWithBrave fixtures to Brave's current <div> markup (the old fixtures had drifted from the live page, which is why the suite stayed green through the breakage).
• Added a dedicated legacy <span> case so back-compat stays covered.
• tests/web-search: 25 tests green; prettier --check clean.

Note on live verification: repeat scrapes from my IP are currently answered with HTTP 429 (Brave rate-limits rapid keyless requests — consistent with the best-effort caveat in #687), so the fixtures mirror the live markup captured in the issue's evidence rather than a fresh scrape.

[tongshu2023]

Apologies — I missed that #688 (opened earlier by the team) already covers this. Closing in favor of that PR. The tests added here are free to cherry-pick if useful.