Integer size for FTorch can now be controlled using parameter
`ftorch_int`.

Currently, the default is set to `int32` but this could be changed if
required.

This fixes an issue on windows with the latest version of ifx/ifort
which builds with a different integer type leading to the following
error:

```
error #6284: There is no matching specific function for this generic function reference.   [TORCH_TENSOR_FROM_ARRAY]
   in_tensors(1) = torch_tensor_from_array(in_data, tensor_layout, torch_kCPU)
-------------------^
```

fixes [#183](#183)

There is an issue when building on mac (arm_64) or windows. The version
of `libtorch` exposes a torch tensors shape (`t->sizes().data()`) as a
`const long long int*` instead of just a `const long int*` like on linux
and mac (x86).

This commit adds preprocessor macro to switch between implementations
automatically detecting the correct version at CMake build stage.

…ccess intel compilers.