feat: add Ollama provider support with streaming inference

[independenter]

Description

This PR adds comprehensive support for the Ollama LLM provider, enabling users to run local models through Ollama with full streaming support.

Changes Made

• ✅ Implemented ollama_text_generation and ollama_chat_generation methods in [server/lib/inference/init.py](server/lib/inference/init.py)
• ✅ Added api_url field to ProviderDetails and Provider entities in [server/lib/entities.py](server/lib/entities.py) for custom API endpoints
• ✅ Registered Ollama routes in [server/app.py](server/app.py)
• ✅ Updated [server/models.json](server/models.json) with Ollama model configurations (e.g., gemma2, llama3)
• ✅ Added comprehensive documentation in Chinese: docs/添加大模型提供商指南.md

Features

• 🔄 Full streaming support for real-time token generation
• 🎯 Support for both text generation and chat modes
• ⚙️ Configurable API URL (default: http://localhost:11434)
• 📊 Comprehensive parameter support (temperature, top_p, top_k, etc.)
• ❌ Graceful cancellation support

How to Test

1. Install Ollama: https://ollama.com/
2. Pull a model: ollama pull llama3 or ollama pull gemma2
3. Ensure Ollama is running (default: http://localhost:11434)
4. Start the OpenPlayground server
5. Select an Ollama model from the dropdown
6. Start chatting and verify streaming responses work correctly

Configuration Example

Add to [server/models.json](server/models.json):

{
  "ollama": {
    "models": {
      "llama3": {
        "enabled": true,
        "status": "ready",
        "parameters": {
          "temperature": 0.7,
          "topP": 0.9,
          "topK": 40,
          "maximumLength": 512
        }
      }
    },
    "requiresAPIKey": false,
    "remoteInference": true,
    "apiURL": "http://localhost:11434"
  }
}