User Preferences for Large Language Model versus Template-Based Explanations of Movie Recommendations: A Pilot Study