MM-BigBench: Evaluating Multimodal Models on Multimodal Content Comprehension Tasks