Fix missing class in cpp tensor documentation (#54488)

Summary:
The given example in the documentation does not compile due to the missing `torch::`. It is correct in the tutorial about [writing a custom extension ](https://pytorch.org/tutorials/advanced/cpp_extension.html#writing-a-mixed-c-cuda-extension)

Pull Request resolved: #54488

Reviewed By: bdhirsh

Differential Revision: D27267000

Pulled By: glaringlee

fbshipit-source-id: 86a46d656c1a4fa4098287a6a43a38d1ef80171e

docs/cpp/source/notes/tensor_basics.rst

@@ -76,7 +76,7 @@ CUDA accessors
 .. code-block:: cpp
 
   __global__ void packed_accessor_kernel(
-      PackedTensorAccessor64<float, 2> foo,
+      torch::PackedTensorAccessor64<float, 2> foo,
       float* trace) {
     int i=threadIdx.x
     gpuAtomicAdd(trace, foo[i][i])