Add support for OpenAI, Cohere, and improve Ollama integration

[maddygoround]

This change extends DeepReview to support multiple Large Language Model providers, including local Ollama instances, OpenAI API, and Cohere API.

Key changes:

• Modified config.yaml structure to support provider-specific configurations (Ollama, OpenAI, Cohere, and a generic API placeholder).
• Added llm_provider setting to select the active LLM provider.
• Updated BranchDiffAnalyzer to load new configurations and initialize the appropriate LLM client (Ollama, ChatOpenAI, Cohere) via a unified get_llm(model_type) method.
• API keys for OpenAI and Cohere can be sourced from config or environment variables (OPENAI_API_KEY, COHERE_API_KEY).
• analyze_file method now routes requests to the correct LLM provider, using chat-style interaction for OpenAI and LLMChain for Ollama/Cohere.
• interactive_qa updated to use the configured QA model from the chosen provider, adapting interaction style (chat vs. LLMChain).
• Command-line arguments enhanced to allow overriding config settings for LLM provider, models, main branch, output options, etc.
• Improved error handling and logging for API calls, configuration loading, and git operations.
• Updated README.md to document new features, command-line arguments, and detailed configuration for each provider.
• Added langchain-community, langchain-openai, and langchain-cohere to setup.py dependencies.
• Initial unit tests were added but encountered environment issues; these can be revisited.