Web Search Plus v2.5.0

Web Search Plus v2.5.0

Reliability + result-quality release.

Reliability

• Enforces remaining time budgets around research/extraction futures so slow providers no longer hold the whole request hostage.
• Preserves partial completed results when one provider or extraction worker is slow.
• Handles 429 Retry-After separately from generic 503 retries and caps inline waiting.
• Decays stale provider cooldown failure history instead of escalating forever.

Search quality

• Adds provider performance memory for adaptive routing.
• Adds spam/mirror result filtering and domain diversity safeguards.
• Avoids lookalike-domain false positives by matching blocked domains only as exact domains or true subdomains.
• Preserves explicit domain intent for site: and include_domains requests.

Credits / PR attribution

• #51 by @robbyczgw-cla — reliability fixes for budgets, futures, 429 handling, and cooldown decay.
• #53 by @robbyczgw-cla — rebased continuation of #52 with adaptive routing and result-quality filtering.

Validation

• python3 -m compileall -q .
• python3 -m pytest tests/ -q → 256 passed
• ruff check .
• git diff --check
• Changed-file privacy scan passed