feat: add Tavily search as configurable alternative to Serper

[tavily-integrations]

Summary

Adds Tavily as an optional, configurable search provider alongside the existing Serper integration (additive/parallel strategy). When configured, search queries are routed to Tavily instead of Serper; otherwise the existing Serper path is unchanged.

Changes

minimax_search_browse.py

• Imported TavilyClient from tavily-python
• Added search_tavily(query) function that calls TavilyClient.search() and normalizes results to match Serper's shape (title, url, snippet)
• Added provider-selection logic in get_search_results(): routes to Tavily when SEARCH_PROVIDER=tavily or TAVILY_API_KEY is present; defaults to Serper otherwise

pyproject.toml

• Added tavily-python>=0.3.0 to dependencies list

Environment Variable Changes

• TAVILY_API_KEY (new, optional) — activates Tavily search path
• SEARCH_PROVIDER (new, optional) — set to tavily or serper; defaults to Serper when unset
• SERPER_API_KEY (existing, unchanged)

Notes for Reviewers

• Tavily's content field is mapped to snippet in normalized results, which get_brief_text() already handles
• Existing Serper code paths are completely untouched
• Provider selection priority: explicit SEARCH_PROVIDER env var > presence of TAVILY_API_KEY > default (Serper)

Automated Review

• Passed after 1 attempt(s)
• Final review: The Tavily additive migration is functionally correct and well-scoped. The search_tavily() function uses the correct SDK patterns, properly normalizes results to the Serper shape expected by get_brief_text(), and the provider-selection logic in get_search_results() is sensible. The tavily-python>=0.3.0 dependency is correctly added to pyproject.toml. Three minor issues exist but none block approval.