(Closes #3259) shorten generated names for basis/diff-basis arrays

[arporter]

No description provided.

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 99.95%. Comparing base (bc9b4db) to head (66ced99).

Additional details and impacted files

@@           Coverage Diff           @@
##           master    #3264   +/-   ##
=======================================
  Coverage   99.95%   99.95%           
=======================================
  Files         377      377           
  Lines       53667    53688   +21     
=======================================
+ Hits        53645    53666   +21     
  Misses         22       22           

☔ 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.

[arporter]

This PR shortens the generated names used for various variables in order to avoid hitting compiler limitations as reported by Tom in #3259. Since we're no longer basing the names on existing variable names (which must be unique) I've had to increase the use of tags to ensure we cope in the face of name collisions.
One for @hiker, @LonelyCat124 or @sergisiso.
I've fired off the integration tests.

[sergisiso]

Looks good @arporter. Its good to see more symbols managed by tags. My only suggestion is to see if we can keep short names still readable, but feel free to ignore if it is to much work.

[arporter]

Thanks for those suggestions.
As discussed, I've added a check on the length of variable names into the Fortran writer. Arbitrarily, this checks for consistency with Fortran2008. I guess we may want the ability to disable this check in future?
Ready for another look now @sergisiso

[sergisiso]

It's looking good @arporter but I have a couple further questions.

[arporter]

Ready for another look now.

[sergisiso]

@arporter Almost there but one more question

[sergisiso]

I was ready to approve but I checked this example (from src/psyclone/tests/lfric_test.py) and the basis_/diff_basis_ /map_/ndf_/undf_ do get shortened to, .e.g: diff_basis_a4_an_or_wh_a_vy_lg_ne_qr
but the proxy and local stencil are still:

• an_operator_with_a_very_long_name_proxy
• an_operator_with_a_very_long_name_local_stencil

I know this is not exactly in the title of the PR, but do you want to also solve those?
And why did the new backend check not complain here?

[arporter]

My feeling is that those are OK as the bulk of the length is due to the original name chosen by the scientist and they have control over that (and I'd like to get on to new things :-) ). The new backend check did not complain because it is generous and follows the F2008 standard of 63 chars.

[sergisiso]

Ah sorry, I didn't realise the limit is 63 and the fs_name max length is set to 21, this is more than enough