- Added getActiveModel() function to automationService for dynamic model selection
- Replaced all hardcoded 'qwen-coder-plus' strings with getActiveModel() calls
- Added localStorage sync when model is changed in orchestrator
- This enables Ollama and other models to work in all automation tasks