refactor(qdp): rename validate_tensor to validate_tensor_cpu and drop redundant check

qdp/qdp-python/src/engine.rs

@@ -16,7 +16,7 @@
 
 use crate::pytorch::{
     extract_cuda_tensor_info, get_torch_cuda_stream_ptr, is_cuda_tensor, is_pytorch_tensor,
-    validate_cuda_tensor_for_encoding, validate_shape, validate_tensor,
+    validate_cuda_tensor_for_encoding, validate_shape, validate_tensor_cpu,
 };
 use crate::tensor::QuantumTensor;
 use numpy::{PyReadonlyArray1, PyReadonlyArray2, PyUntypedArrayMethods};
@@ -206,7 +206,7 @@ impl QdpEngine {
         }
 
         // CPU tensor path
-        validate_tensor(data)?;
+        validate_tensor_cpu(data)?;
         // PERF: Avoid Tensor -> Python list -> Vec deep copies.
         //
         // For CPU tensors, `tensor.detach().numpy()` returns a NumPy view that shares the same

qdp/qdp-python/src/pytorch.rs

@@ -33,12 +33,12 @@ pub fn is_pytorch_tensor(obj: &Bound<'_, PyAny>) -> PyResult<bool> {
     Ok(module_name == "torch")
 }
 
-/// Helper to validate CPU tensor
-pub fn validate_tensor(tensor: &Bound<'_, PyAny>) -> PyResult<()> {
-    if !is_pytorch_tensor(tensor)? {
-        return Err(PyRuntimeError::new_err("Object is not a PyTorch Tensor"));
-    }
-
+/// Validate that a PyTorch tensor lives on the CPU.
+///
+/// The caller must have already confirmed `tensor` is a `torch.Tensor` (e.g.
+/// via [`is_pytorch_tensor`]). This function only checks device placement, so
+/// it also rejects non-CPU backends such as CUDA, MPS, XLA, and HPU.
+pub fn validate_tensor_cpu(tensor: &Bound<'_, PyAny>) -> PyResult<()> {
     let device = tensor.getattr("device")?;
     let device_type: String = device.getattr("type")?.extract()?;