Operator overloads not working correctly for non-integer scalars

[jwallwork23]

Unfortunately, there's an oversight in my implementation of the operator overloads that involve scalars. They only seem to currently work for integer scalars (only integer scalars are considered in the autograd demo).

For example, changing the divisor with

diff --git a/examples/6_Autograd/autograd.f90 b/examples/6_Autograd/autograd.f90
index cf60494..9d73853 100644
--- a/examples/6_Autograd/autograd.f90
+++ b/examples/6_Autograd/autograd.f90
@@ -52,7 +52,7 @@ program example
   ! Check arithmetic operations work for torch_tensors
   write (*,*) "a = ", in_data1(:,1)
   write (*,*) "b = ", in_data2(:,1)
-  Q = 3 * (a**3 - b * b / 3)
+  Q = 3 * (a**3 - b * b / 3.0)
 
   ! Extract a Fortran array from a Torch tensor
   call torch_tensor_to_array(Q, out_data, shape(in_data1))

on the main branch results in the test failure:

test 2
    Start 2: fautograd

2: Test command: /home/joe/software/FTorch/src/build/test/examples/6_Autograd/autograd
2: Test timeout computed to be: 1500
2:  a =    2.00000000       3.00000000    
2:  b =    6.00000000       4.00000000    
2:  Q = 3 * (a ** 3 - b * b / 2) =   24.0000000       81.0000000    
2: FAILED :: [torch_tensor_to_array] relative tolerance =  0.1000E-04
2:  Error :: out_data does not match expected value
2: STOP 999
2/2 Test #2: fautograd ........................***Failed  Required regular expression not found. Regex=[Autograd example ran successfully
]  0.23 sec