Implement DeepSeek-V3 LLM Provider Driver

[hardikkaurani]

🟡 Intermediate Issue

Problem

EvalForge currently supports OpenAI, Anthropic Claude, Google Gemini, and Ollama providers but does not have a DeepSeek driver.

Why It Matters

DeepSeek-V3 and DeepSeek-R1 are increasingly popular open-weight models. First-class provider support expands EvalForge's evaluation coverage.

Suggested Files

• backend/app/evaluation/providers/base.py (review interface)
• backend/app/evaluation/providers/deepseek_provider.py (new file)

Definition of Done

• DeepSeekLLMProvider class inheriting from BaseLLMProvider created.
• Provider registered in the provider factory.
• Unit tests with mocked API responses added.

[hardikkaurani]

👋 Maintainer Solution Guide:

1. Review the base provider interface in backend/app/evaluation/providers/base.py
2. Create deepseek_provider.py implementing BaseLLMProvider with generate() and score() methods
3. DeepSeek uses an OpenAI-compatible API at https://api.deepseek.com/v1/chat/completions
4. Register in the provider factory so provider='deepseek' works in evaluation config
5. Add unit tests mocking the DeepSeek API response

Exciting new feature — expand model coverage! 🤖