Fix issue #50: Add a new speaker to the panel

[chuanli11]

This pull request fixes #50.

The issue has been successfully resolved. The changes comprehensively integrate the DeepSeek-Llama3.3-70B model from Lambda's inference API into the panel discussion system:

Core Integration Changes:

• Added a new LambdaClient class that implements the OpenAI-compatible API format for Lambda's endpoint (https://api.lambda.ai/v1)
• Updated the main application to include DeepSeek as the 4th panelist alongside GPT-5, Claude, and Gemini
• Modified the TurnManager to include "deepseek" in the panelist rotation
• Added Lambda API key configuration to the environment template and config loading

Infrastructure Updates:

• Added requests dependency for HTTP API calls to Lambda
• Updated documentation to reflect the new 4-panelist setup
• Added comprehensive error handling for API failures and missing dependencies

Testing Coverage:

• Added unit tests for Lambda client initialization and response generation
• Added integration tests verifying DeepSeek participates properly in discussion flow
• Updated existing tests to account for the 4th panelist

The implementation follows the same patterns as the existing LLM clients, uses proper async/await patterns, includes retry logic with backoff, and maintains consistency with the application's architecture. The DeepSeek model will now participate as an equal panelist in all discussion rounds, contributing its perspective alongside the other AI models.

Automatic fix generated by OpenHands 🙌