Enhancing Large Vision-Language Models with Layout Modality for Table Question Answering on Japanese Annual Securities Reports