Fix constituent handling

[peverwhee]

Main fix:

• physics/utils/physics_data.F90: update the logic to not return after determining if a variable is a constituent. This keeps the "initialized" logic so that SIMA doesn't try to read in constituents that are initialized by the SE dycore.
  • Basically, only set find_input_name_idx to "const_idx" (the parameter that determines if a variable is a constituent to be read in from file) if it's not marked as initialized or protected.

Smaller fixes

• dynamics/se/namelist_definition_se_dycore.xml: add namelist default for ne3 grid
• data/write_init_files.py: trim constituent standard names output to log

[nusbaume]

Thanks for tracking down this bug! I just had one (optional) code clean-up request.

Also, an FYI that I have added instructions for the current CAM-SIMA regression tests here:
https://github.com/ESCOMP/CAM-SIMA/wiki/CAM-SIMA-testing#regression-testing-manual

Please let me know if you have any questions!

[peverwhee]

tests run and passed through model_build:
SMS_Ln2.ne3pg3_ne3pg3_mg37.FPHYStest.derecho_intel.cam-outfrq_kessler_derecho_nooutput
SMS_Ln2.mpasa480_mpasa480.FKESSLER.derecho_gnu.cam-outfrq_kessler_mpas_derecho_nooutput
SMS_Ln2.ne3pg3_ne3pg3_mg37.FPHYStest.derecho_gnu.cam-outfrq_held_suarez_derecho_nooutput

[mwaxmonsky]

Looks great @peverwhee!

[peverwhee]

Requesting re-review for new fix:

• control/cam_comp.F90: move call to stepon_run1 to cam_timestep_init so that host variables are initialized before any call to CCPP physics. Also rename stepon_run1 to stepon_timestep_init
• dynamics/*/stepon.F90: rename stepon_run1 to stepon_timestep_init

All regression tests pass.