(closes #3052) Fix missing psylayer openmp symbols

[sergisiso]

EPCC reported that combining everything mentioned in the ticket the "th_idx" and "omp_get_thread_num" declarations are missing. I added a similar test.

Both have in common that are declared on self. scope, if I move their declaration to self.ancestor(Routine).scope the problem is gone. EPCC has verified that this works for them. Other symbols declared on self.scope are fine.

[codecov]

Codecov Report

All modified and coverable lines are covered by tests ✅

Project coverage is 99.92%. Comparing base (c16ab26) to head (d19ad0d).
Report is 8 commits behind head on master.

Additional details and impacted files

@@           Coverage Diff           @@
##           master    #3053   +/-   ##
=======================================
  Coverage   99.92%   99.92%           
=======================================
  Files         366      366           
  Lines       51677    51678    +1     
=======================================
+ Hits        51637    51638    +1     
  Misses         40       40           

☔ 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]

Thanks for this Sergi. It would be good to take a bit of time here to figure out what is actually going wrong (or at least, as much detail as you can provide) so that we can fix it properly in slower time. It would also be good to add compilation testing to the new test if possible.
I'll launch the integration tests in the meantime (unless you have already done so).

[sergisiso]

@arporter This is ready for another review

[sergisiso]

Jaffery confirmed that the new fix still works for EPCC profiling

[arporter]

Well done on tracking this down.
I've just set the integration tests going again as this is an important change (although I'm not sure how much of psy_data_node they touch really). If they are OK then this is good to go.

[sergisiso]

There is a nemov5 file that failed, but it is not related to this PR (has not psy_data or lfric), do you want me to attempt filtering it here or proceed and deal with it in a NEMO PR?

[arporter]

I'll merge this one and then we can fix it separately I think.