Add support for OpenAI o3 mini and GPT-4o models

[bhouston]

Description\nCurrently, MyCoder uses Anthropic's Claude model via Vercel AI SDK. We should expand model support to include OpenAI's models, specifically o3 mini and GPT-4o.\n\n## Why\n- Provide more model options for users\n- o3 mini offers a cost-effective option for simpler tasks\n- GPT-4o is OpenAI's latest multimodal model with strong capabilities\n\n## Implementation Plan\n1. Add OpenAI model configuration using the existing package (already installed)\n2. Update the agent configuration to support model selection\n3. Implement model selection in the CLI interface\n4. Add appropriate documentation\n\n## Technical Details\nThe project already has the necessary dependencies:\n- is already installed in the agent package\n\nSimilar to how the Anthropic Claude model is configured in , we can add support for OpenAI models:\n\n\n\n## Testing\n- Test with both models for different types of tasks\n- Compare performance, cost, and capabilities with Claude models\n- Ensure all existing functionality works with the new models

[bhouston]

Additional Implementation Details\n\n- We should leverage the existing CLI config capabilities to allow users to specify the model via command line arguments (e.g., or ).\n- Add support for setting the default model in the persistent configuration.\n- For authentication, we'll use the standard OpenAI API key environment variable (OPENAI_API_KEY), similar to how we handle Anthropic authentication.\n- This approach will maintain consistency with the current design while expanding model options.