- Add _is_ollama_model method to detect Ollama models consistently
- Skip response_format validation for Ollama models in _validate_call_params
- Filter out response_format parameter for Ollama models in _prepare_completion_params
- Add comprehensive tests for Ollama response_format handling
- Maintain backward compatibility for other LLM providers
Fixes#3082
Co-Authored-By: João <joao@crewai.com>