(WIP) Save 3D marine analysis

[Dooruk]

I have this version that works for the marine side as we won't have 4D analyses fields for a while.

For geos_cf and geos_atmos there is additional complexity in terms of:

• use of model/r2d2.py by defining an dictionary for r2d2 to refer.
• What type of grid will be stored? latlon or cubed?
• (minor) is horizontal res part of r2d2 analysis item?
• 4D analysis (not handled in this PR).

So I wanted to check with you @jeromebarre and @mer-a-o, whether you like this approach and would like to build on this, or would like to have a completely different geos_cf saveanalysis task? Sorry, I dragged this for a while, wanted SaveForecast to go in first.

For the last point (4D fields), I have

swell/src/swell/utilities/geos.py

Lines 423 to 506 in e3007e1

	# --------------------------------------------------------------------------------------------------
	def states_generator(
	self,
	background_frequency: str,
	window_length: str,
	window_begin_iso: str,
	model: str = 'geos_marine',
	marine_models: list = []
	) -> list:
	"""
	Generates a list of states for JEDI configuration based on the model type.
	Args:
	background_frequency (str): ISO 8601 duration between background states.
	window_length (str): ISO 8601 duration of the DA window.
	window_begin_iso (str): ISO 8601 datetime string for the start of the window.
	model (str): Model component name (e.g., 'geos_marine'). Defaults to 'geos_marine'.
	marine_models (list): List of active marine model components (e.g., ['cice6']).
	Returns:
	list: A list of dictionaries, each representing a state with its date and filenames.
	"""
	states = []
	self.logger.info('Generating states for model: '+model)
	if model in ("geos_marine"):
	states = self.marine_states(background_frequency, window_length, window_begin_iso,
	marine_models)
	return states
	# --------------------------------------------------------------------------------------------------
	def marine_states(self,
	background_frequency: str,
	window_length: str,
	window_begin_iso: str,
	marine_models: list
	) -> list:
	"""
	Generates states specifically for the geos_marine model component.
	Args:
	background_frequency (str): ISO 8601 duration between background states.
	window_length (str): ISO 8601 duration of the DA window.
	window_begin_iso (str): ISO 8601 datetime string for the start of the window.
	marine_models (list): List of active marine model components.
	Returns:
	list: List of dictionaries containing JEDI state information for marine models.
	"""
	static_part = {"basename": "./", "read_from_file": 1}
	# Calculate the number of states using background frequency and window length
	number_of_states = int(isodate.parse_duration(window_length)
	/ isodate.parse_duration(background_frequency)) + 1
	self.logger.info('Number of states: ' + str(number_of_states-1))
	# Generate the list of states dictionary with date and marine filename entries.
	# The date is calculated by adding the background frequency to the window begin date.
	# The ocn_filename is calculated by adding the background frequency to the window begin date
	states = []
	# For FGAT and 4D-Var, the first state is the background state, hence we need to
	# skip the first state in the loop by adding "-1" to the range function.
	for i in range(1, number_of_states):
	hours = int((isodate.parse_duration(background_frequency) * i).total_seconds() / 3600)
	state_dto = isodate.parse_datetime(window_begin_iso) \
	+ isodate.parse_duration(background_frequency) * i
	state = {
	"date": state_dto.strftime(datetime_formats['iso_format']),
	"ocn_filename": "ocn.fc." + window_begin_iso + "." + f"PT{hours}H" + ".nc"
	}
	if 'cice6' in marine_models:
	state.update({"ice_filename": "ice.fc." + window_begin_iso + "." + f"PT{hours}H" +
	".nc"})
	state.update(static_part)
	states.append(state)
	return states
	# --------------------------------------------------------------------------------------------------

which may be useful to define analyses using background frequency but your call.