Text this: Towards a Large Language-Vision Question Answering Model for MSTAR Automatic Target Recognition