Unified Large Language Models for Misinformation Detection in Low-Resource Linguistic Settings