[WIP]Fix PGI runtime issue with class(*)

[nikizadehgfdl]

• Some tests such as global_ALE_z crash under PGI (ncrc4.pgi20 or ncrc5.pgi227) with
  FATAL from PE 27: unsupported attribute type: get_variable_attribute_0d: file:INPUT/tideamp.nc- variable:GRID_X_T attribute: axis
• PGI in general has issues with class(*) construct and in this case
  cannot recognize the axis argument to be a string.
• This mod helps PGI recognize that the argument is a string.

[codecov]

Codecov Report

Merging #346 (b4f1f0a) into dev/gfdl (b075794) will not change coverage.
The diff coverage is n/a.

❗ Current head b4f1f0a differs from pull request most recent head f8d215a. Consider uploading reports for the commit f8d215a to get more accurate results

@@            Coverage Diff            @@
##           dev/gfdl     #346   +/-   ##
=========================================
  Coverage     38.38%   38.38%           
=========================================
  Files           268      268           
  Lines         76010    76010           
  Branches      13987    13987           
=========================================
  Hits          29176    29176           
  Misses        41595    41595           
  Partials       5239     5239           

see 1 file with indirect coverage changes

📣 We’re building smart automated test selection to slash your CI/CD build times. Learn more

[marshallward]

Is this still happening in the current dev/gfdl? At least one instance of tideamp.nc was moved off of FMS and to the netCDF layer to avoid this issue.

[marshallward]

@nikizadehgfdl Is there any update on this? Is this still a problem on the latest branch?

[marshallward]

@nikizadehgfdl I tried the latest dev/gfdl with PGI and global_ALE/z appeared to run without any regressions, so I am tempted to close this issue.

Having said that, I did find a problem with data_override in PGI for several of the ice-ocean experiments. So I don't think we're at the end of this problem.

I'll let you decide how to proceed with this.

[marshallward]

I suspect this is no longer necessary but I suppose there is no harm in it.

[marshallward]

Gaea regression: https://gitlab.gfdl.noaa.gov/ogrp/MOM6/-/pipelines/19482 ✔️