[PSyData] Minor extraction library updates

[hiker]

With the kernel extraction started to be used more (e.g. in NG-ARCH), I am in the process of integration this with our Baf build system (which is the lfric Fab build system). That will involve copying the PSyclone NetCDF LFRic extraction files into the build system (so they can be injected into the Fab build, so no need for creating libraries etc). A few changes required:

• Minor updates to debug output (using trim to remove white spaces)
• LFRic's real field types have been renamed (field_r{{prec}}_mod to field_real{{prec}}_mod etc). We can't simply switch this, since atm our infrastructure is outdated (see LFRic changed name of field_r*types #2623 for details, and also PSyclone is now unfriendly to case-insensitive filesystems #2239 - best solution is to switch to the now stand-alone lfric infrastructure files, which should support non-MPI builds now). As a quick solution till this is properly fixed, I'll add a second copy of the jinja files with the new names.
• The count of errors is not correct, the compare library should be using (I think)

      all_results(current_index, COUNT_NEG_3) = COUNT(tmp > 1.0D-3)
      all_results(current_index, COUNT_NEG_6) = COUNT(tmp > 1.0D-6) - all_results(current_Index, COUNT_NEG_3)
      all_results(current_index, COUNT_NEG_9) = COUNT(tmp > 1.0D-9) - all_results(current_Index, COUNT_NEG_6) &
                                                                    - all_results(current_Index, COUNT_NEG_3)

• We should also include the overall number of elements (and/or the number of elements >1.0d-3 error)