Text this: Large multimodal model for open vocabulary semantic segmentation of remote sensing images