Add Ollama support for local models

[xmaks82]

Description

Add Ollama as a local LLM provider, allowing users to run models locally without any API key.

Implementation

1. Create src/agent/clients/ollama_client.py extending BaseLLMClient
2. Implement chat(), stream(), chat_with_tools() using Ollama's OpenAI-compatible API
3. Add model mappings to src/agent/clients/factory.py
4. Default endpoint: http://localhost:11434

Reference

• Ollama API docs
• Existing client example: src/agent/clients/groq_client.py

This is a great first contribution — the pattern is well-established in the codebase.