Use get_device_index method in test

[jwallwork23]

Looks like I failed to re-test later changes in #222.

Building with CUDA support I got

/home/joewa/software/FTorch/src/build/test/unit/test_tensor_interrogation_cuda.F90:64:19:

   64 |     if (expected /= torch_tensor_get_device_index(tensor)) then
      |                   1
Error: Function ‘torch_tensor_get_device_index’ at (1) has no IMPLICIT type
/home/joewa/software/FTorch/src/build/test/unit/test_tensor_interrogation_cuda.F90:77:8:

   77 |    use test_tensor_interrogation_cuda
      |        1
Fatal Error: Cannot open module file ‘test_tensor_interrogation_cuda.mod’ for reading at (1): No such file or directory
compilation terminated.

which is fixed here.

[jatkinson1000]

Thanks @jwallwork23

Another good reason for getting a GPU runner going on occasion - @dorchard