MALoRA: Mixture of Asymmetric Low-Rank Adaptation for Enhanced Multi-Task Learning