Text this: KOMPSAT-3/3A Image-text Dataset for Training Large Multimodal Models