Implement OpenAI provider support

[thomas-vilte]

Problem Statement

I have observed that users currently lack the ability to utilize OpenAI's models for generating commit messages within MateCommit. As we aim to support a diverse range of AI backends, the absence of an OpenAI integration limits the tool's utility for developers who rely on GPT-3.5 or GPT-4 for high-quality text generation. Furthermore, with the recent introduction of the new provider interface, we need to ensure all new integrations adhere to this standard for maintainability.

Proposed Solution

I propose implementing a dedicated OpenAI provider that fully integrates with our new provider interface. The implementation should include the following steps:

1. Create a new OpenAIProvider struct that satisfies the provider interface contract.
2. Implement the logic to communicate with the OpenAI API endpoints.
3. Add configuration support to handle the OpenAI API key securely.
4. Allow users to configure the specific model they wish to use (e.g., gpt-3.5-turbo, gpt-4), with a sensible default.

Alternatives Considered

I considered relying solely on local LLMs or other existing providers, but OpenAI remains a standard in the industry with high availability and quality. I also considered implementing this using the legacy provider pattern, but that would introduce technical debt immediately; therefore, using the new interface is mandatory.

Additional Context

This implementation should serve as a reference example for future cloud-based provider integrations. Please ensure error handling covers cases like quota limits and invalid API keys.