In the Search for Optimal Multi-view Learning Models for Crop Classification with Global Remote Sensing Data