Add Tavily option for deeptracehub-search-crawl

[evanYDL]

Search Provider Migration: deeptracehub-search-crawl

Adds Tavily as an additional option/parallel path while preserving existing provider behavior.

Introduces Tavily, a search API optimized for LLM applications.

Changes

All changes are complete and consistent. Here's a summary of everything that was implemented for the deeptracehub-search-crawl migration unit:

Summary of Changes

All files already had the core Tavily integration in place. The one remaining item was:

config/deepseek.yaml (updated)

• Added # options: jina, serper.dev, local, tavily comment to crawl_tool (line 41)
• Added # options: serper.dev, empty, local, tavily comment to search_tool (line 56)

Previously completed changes (verified as correct)

File	Change
tools/types.py	TAVILY = "tavily" added to both SerpProvider and WebProvider enums
tools/search.py	TavilySearchTool(SearchToolBase) added — uses AsyncTavilyClient().search(), maps results to SerpEntry with SerpProvider.TAVILY, integrates with SearchCache and execute_with_retries
tools/search.py	TavilyWebCrawlTool(WebCrawlToolBase) added — uses AsyncTavilyClient().extract(urls=[url]), maps results to WebContentV2 with WebProvider.TAVILY, integrates with WebCrawlCache and execute_with_retries
src/agents/base.py	TavilySearchTool and TavilyWebCrawlTool imported and registered in TOOL_CLASS_SEARCH / TOOL_CLASS_CRAWL lists
requirements.txt	tavily-python==0.5.0 present
.env	TAVILY_API_KEY="" present alongside existing SERPERDEV_TOKEN and JINA_TOKEN
config/redsearcher.yaml	tavily documented as an option for both search_tool and crawl_tool

Migration strategy: ADDITIVE

• All existing providers (Serper.dev, Jina, Local) are preserved unchanged
• Tavily is available as a parallel option, selectable via config (search_tool: tavily / crawl_tool: tavily)
• No existing dependencies or env vars were removed