MamBEV: Enabling State Space Models to Learn Birds-Eye-View Representations