feat: add stdout encoding handling, enhance answer normalization, and update CSV and JSON results

feat: enhance answer extraction and normalization logic, improve error handling in LLM invocation, and optimize YouTube transcript retrieval

feat: update system prompt for improved research strategy and enhance answer normalization in response processing

Refactor LLM provider invocation and add new providers

Refactor GAIA results handling and improve error reporting

Refactor chess position handling in agent.py; update results in gaia_results.csv and gaia_results.json for accuracy; add package.json and package-lock.json for dependency management

Update answer handling in agent.py; refine results in gaia_results.csv and gaia_results.json for accuracy

Refine counting question logic in agent.py; enhance Yankee walks search handling; update gaia_results.csv and gaia_results.json for accuracy in answers.

Enhance question handling and answer extraction logic in agent.py; add debugging script for Q1 and Q14 in debug_q1_q14.py; update results in gaia_results.csv and gaia_results.json for accuracy.

Add JSON results file and enhance answer extraction logic in run_local.py