Bring LiteLLM support

[amanjaiswal73892]

Add support for LiteLLM in the llm/response_api.py for better extensibility and maintainability of newer LLMs.

• Add new Model class for LiteLLM similar to ChatCompletionModel.
• Include support for anthropic related parameters (cache_breakpoints etc).
• Include other parameters like (reasoning effort) etc.
• Add cost tracking in tracking.py for litellm and use litellm helper functions.
• Write tests using 'mock' param of liteLMM completion.
• Test function calling support.
• Replicate OAI/Anthropic Agents using LiteLLM and compare outputs.