[FEATURE] crawl4ai to perform extensive scraping functionality

current agent web-search uses duckduckgo library. fast and lightweight, but has some limitations:

• prone to rate limiting on repeated requests
• returns a limited result content

to improve reliability and search depth, crawl4ai as an additional scraping layer.

new behavior:

• duckduckgo remains the default for quick lookups
• if results are empty or more depth is needed, crawl4ai is used as a fallback

crawl4ai supports multiple search engines, retry logic, and stealth scraping (e.g., anti-captcha). this hybrid setup keeps performance fast while improving coverage and resilience for more demanding and in-depth use cases.