feat: implement Ollama provider for LLM abstraction

[bhouston]

Implement Ollama Provider

This PR adds support for the Ollama API to the MyCoder LLM abstraction. Ollama is an open-source framework that allows running large language models locally.

Implementation Details:

• Created an Ollama provider implementation in packages/agent/src/core/llm/providers/ollama.ts
• Supports configurable endpoint URL (default to http://localhost:11434)
• Supports the chat API endpoint with proper message formatting
• Handles tool/function calling
• Registered the provider in the LLM provider registry

Configuration:

• Default endpoint: http://localhost:11434
• Can be configured via:
  • baseUrl option when creating the provider
  • OLLAMA_BASE_URL environment variable

Usage Example:

import { createProvider } from './core/llm/provider.js';

const provider = createProvider('ollama', 'llama3.2', {
  baseUrl: 'http://localhost:11434',
});

const response = await provider.generateText({
  messages: [
    { role: 'user', content: 'Hello, how are you?' }
  ],
  temperature: 0.7,
});

Fixes #209