Allow APP to differ between RUNs

[DavidHuber-NOAA]

Description

This enables APP to be specified for each RUN. This also removes the need for a _no_run configuration dictionary and somewhat simplifies the _init_finalize method.

Resolves #2908
Resolves #2956

Type of change

• New feature (adds functionality)

Change characteristics

• Is this a breaking change (a change in existing functionality)? NO
• Does this change require a documentation update? NO
• Does this change require an update to any submodules? NO

How has this been tested?

• Generated .xml files on Hera for all available CI tests and verified they were identical to develop.
• Created an arbitrary APP configuration ({"gdas": "S2S", "gfs": "ATM"})

Checklist

• My code follows the style guidelines of this project
• I have performed a self-review of my own code
• I have commented my code, particularly in hard-to-understand areas
• My changes generate no new warnings
• New and existing tests pass with my changes

[DavidHuber-NOAA]

This is going to require some more leg work as some APP-specific flags are currently read in without specifying a RUN:

global-workflow/workflow/applications/applications.py

Lines 34 to 50 in c6e3262

	base = conf.parse_config('config.base')
	self.mode = base['MODE']
	if self.mode not in self.VALID_MODES:
	raise NotImplementedError(f'{self.mode} is not a valid application mode.\n'
	f'Valid application modes are:\n'
	f'{", ".join(self.VALID_MODES)}\n')
	self.net = base['NET']
	self.model_app = base.get('APP', 'ATM')
	self.do_atm = base.get('DO_ATM', True)
	self.do_wave = base.get('DO_WAVE', False)
	self.do_wave_bnd = base.get('DOBNDPNT_WAVE', False)
	self.do_ocean = base.get('DO_OCN', False)
	self.do_ice = base.get('DO_ICE', False)
	self.do_aero = base.get('DO_AERO', False)

This will require reworking the get_task_names methods to key off of RUN-specific flags.

[DavidHuber-NOAA]

This PR now correctly generates identical CI XMLs as compared to develop (with the exception of the stage_ic and cleanup jobs which are now run on service partitions). An additional case was tested based on the C96C48_hybatmaerosnowDA case, but the APP was set to ATMA for RUN==gdas and ATM for RUN==gfs. In this case, the gfs tasks in the resultant XML did not include any aero-component jobs.

Marking ready for review.

[emcbot]

Experiment C48_ATM FAILED on Orion in Build# 1 in
/work2/noaa/stmp/CI/ORION/2943/RUNTESTS/EXPDIR/C48_ATM_1975b2f3

[emcbot]

Experiment C96_S2SWA_gefs_replay_ics FAILED on Orion in Build# 1 in
/work2/noaa/stmp/CI/ORION/2943/RUNTESTS/EXPDIR/C96_S2SWA_gefs_replay_ics_1975b2f3

[emcbot]

CI Failed on Orion in Build# 1
Built and ran in directory /work2/noaa/stmp/CI/ORION/2943

Experiment C96C48_hybatmDA_1975b2f3 Terminated with 0 tasks failed and 0 dead at Thu Oct 17 09:31:44 AM CDT 2024
Experiment C96C48_hybatmDA_1975b2f3 Terminated: *STALLED*
Experiment C96_atm3DVar_1975b2f3 Terminated with 0 tasks failed and 0 dead at Thu Oct 17 09:37:53 AM CDT 2024
Experiment C96_atm3DVar_1975b2f3 Terminated: *STALLED*
Experiment C48_ATM_1975b2f3 Terminated with 0
FAIL
FAIL tasks failed and 1 dead at Thu Oct 17 11:28:20 AM CDT 2024
Experiment C48_ATM_1975b2f3 Terminated: *FAIL*
Error logs:
/work2/noaa/stmp/CI/ORION/2943/RUNTESTS/COMROOT/C48_ATM_1975b2f3/logs/2021032312/gfs_metpg2o1.log
Experiment C96_S2SWA_gefs_replay_ics_1975b2f3 Terminated with 0 tasks failed and 0 dead at Thu Oct 17 11:34:29 AM CDT 2024
Experiment C96_S2SWA_gefs_replay_ics_1975b2f3 Terminated: *STALLED*
Experiment C48_S2SW_1975b2f3 Completed 1 Cycles: *SUCCESS* at Thu Oct 17 01:45:14 PM CDT 2024
Experiment C48_S2SWA_gefs_1975b2f3 Completed 1 Cycles: *SUCCESS* at Thu Oct 17 02:46:52 PM CDT 2024

[DavidHuber-NOAA]

I'm having some issues with this branch after merging the interval PR. Converting to draft until I get it squared away.

[DavidHuber-NOAA]

All features are working again. Reopening this PR for review.

[WalterKolczynski-NOAA]

Why is there no DO_AERO for GEFS anymore (and not exported for GFS)?

[DavidHuber-NOAA]

Because aerosols are controlled at the forecast and analysis level, not the application level.

For instance, previously, this was implemented in the config.fcst config by reading the AERO_FCST_RUN variable:

global-workflow/parm/config/gfs/config.fcst

Lines 11 to 21 in 152bb45

	# Turn off waves if not used for this RUN
	case ${WAVE_RUN} in
	both | "${RUN/enkf}" ) ;; # Don't change
	*) DO_WAVE="NO" ;; # Turn waves off
	esac
	# Turn off aerosols if not used for this RUN
	case ${AERO_FCST_RUN} in
	both | "${RUN/enkf}" ) ;; # Don't change
	*) DO_AERO="NO" ;; # Turn aerosols off
	esac

But now, it can rely on just the DO_AERO_FCST variable.

Keeping the DO_AERO flag would be redundant since it is superseded by DO_AERO_FCST (or DO_AERO_ANL in the analysis jobs).

[DavidHuber-NOAA]

I debated whether to add DO_AERO (without exporting) to config.gefs. When analyses are enabled, having the flag will be handy.