Remove deprecated MPS backend support, eliminate torch.mps availability checks

src/config.py

@@ -69,10 +69,6 @@ def get_device_info() -> tuple[str, str]:
             pass
         return device_name, f"CUDA (GPU: {gpu_name})"
     
-    elif torch.backends.mps.is_available():
-        device_name = "mps"
-        return device_name, "Apple Silicon MPS"
-    
     else:
         # CPU fallback - check available RAM
         try: