Extend FTorch to cover 5-dimensional tensors.

[jatkinson1000]

Closes #172

We have had a request for 5D tensor support.
This is from the ConvLSTM: https://paperswithcode.com/method/convlstm which is a type of NN used to analyse gridded data.
Given our initial application (climate sciences) it would make sense to support this as standard.

I have extended the ranks argument in the fypp code to now support 5-D (rank 5) tensors.

One point to discuss is whether we want to support slightly higher in anticipation of other applications (Rank 6 or perhaps 7) bearing in mind that doing so adds quite a lot of code/functions in the f90 file.

We should also perhaps add documentation noting the limit in Rank, and perhaps describing how to extend it for specific cases.

[jatkinson1000]

As discussed over in #166 The tests for autograd were producing an error after this change, despite not touching them.

This could be fixed locally on my machine by checking for (.not. associated(out_data)) rather than the shape of unassociated/unallocated out_data.

However, the code is still failing the CI test here.
My guess would be that this is perhaps due to a regex mismatch that is being dealt with over in #166
So question is whether to:

• merge as is here
• turn autograd test off for now until fixed in Use assertions rather than print statements in tests #166 as not currently an active feature for users

Update: See comment on #166 #166 (comment) and associated PR #174

This seems unrelated to the changes I have made here, so my suggestion is perhaps to turn the autograd test off for now and let this issue be resolved by #166 then turned back on.

[jatkinson1000]

I have just removed the commit that worked around the pointer issues that will be addressed by #175

This should now be merged after:

• change to_array routines to fix pointer issues #175
• Use assertions rather than print statements in tests #166

[jatkinson1000]

Rebased after #175 merged to main.
@TomMelt's changes to regex matching in autograd test as part of #175 mean that #166 is no longer a blocker so this is ready for review.

[jwallwork23]

Single character change PR, love it!