Fix array_strides with nvhpc; Note: not working on Linux_aarch64

[wdeconinck]

Try to fix array_strides computation for NVHPC.
Note, this is not working on Linux_aarch64.

[codecov-commenter]

Codecov Report

❌ Patch coverage is 60.00000% with 28 lines in your changes missing coverage. Please review.
✅ Project coverage is 52.15%. Comparing base (67dd487) to head (55d0afc).
⚠️ Report is 1 commits behind head on develop.

Files with missing lines	Patch %	Lines
src/fckit/fctest.F90	51.72%	28 Missing ⚠️

Additional details and impacted files

@@             Coverage Diff             @@
##           develop      #74      +/-   ##
===========================================
+ Coverage    51.35%   52.15%   +0.80%     
===========================================
  Files           58       58              
  Lines         4812     4880      +68     
  Branches       449      448       -1     
===========================================
+ Hits          2471     2545      +74     
+ Misses        2341     2335       -6     

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.

[wdeconinck]

This is causing some unexpected error at the moment with nvidia/22.11 on the HPC CI, so should not be merged yet until understood.

  + srun -n 4 -c 1 /etc/ecmwf/ssd/ssd1/tmpdirs/***.30495790/atlas/build/src/tests/functionspace/atlas_fctest_functionspace
  srun: Step created for StepId=30495790.223
            827
   before: name = StructuredColumns
   before: owners =             2
   after: name = StructuredColumns
   after: owners =             3
   before: name = PointCloud
   before: owners =             2
   after: name = PointCloud
   after: owners =             3
  /etc/ecmwf/ssd/ssd1/tmpdirs/***.30495790/atlas/src/tests/functionspace/fctest_functionspace.F90:1039: warning: FCTEST_CHECK_CLOSE(field_values_new(k, i), field_values_save(k, i), tol)
   --> [    45.00000000000000      !=    0.000000000000000      ]

[wdeconinck]

This is causing some unexpected error at the moment with nvidia/22.11 on the HPC CI, so should not be merged yet until understood.

A workaround in atlas has been found and merged in develop (ecmwf/atlas#332) that makes this introduction neutral, fixing the failing test.

[wdeconinck]

There are other unforeseen downstream problems unless the entire downstream stack is compiled with this flag.
Bug reports have been submitted to NVHPC compiler developers.