CoMPaSS:_Enhancing_Spatial_Understanding_in_Text-to-Image_Diffusion_Models