- "archive_internal_dir" must be an empty string for top-level archive
   files
 - Rename "file_type" argument to "file_fmt"
 - Rename "external_model" argument to more generic/correct "data_type"
 - If only some files can not be un-zipped, do not fail. We'll handle
   this more gracefully later if it needs to be an error
 - Remove default location logic for now; couldn't get it working
 - Re-add explicit disk path requirement; may re-visit later
 - For fatal error, do not imply that script continues
 - Remove debug prints

"rap_e" files in addition to "rap".

specifically HPSS fails with some kind of logging error. Needs more
investigation.

This commit removes some un-needed, un-initialized variables, adds
modulefiles for the new task on each platform, and adds some unit tests
for retrieve_data.py to test new capabilities.

The new task entry has been added to parm/wflow/coldstart.yaml for
testing, but it should be removed prior to merging this PR since the
task is not integrated with the rest of the workflow yet.

… tasks going forward. Clearly the boundary conditions tasks will need to eventually change also.

 - Link obs needed for task_process_bufr
 - Put obs in $DATA/obs rather than $DATA/gsi
 - Put in logic for getting "rap_e" files at 00 and 12z cycles; I hope
we can extract this logic into the config file later with Jinja
 - Add "task_process_bufr" entry to warmstart.yaml with a dependency on
task get_da_obs
 - Add "CYCLE_TYPE" variable to config_defaults
 - Start a new config.rrfs.yaml file

 - Set machine-specific paths in machine files, this should have been
obvious...
 - Use RRFS_NA_13km domain for RRFS
 - Re-organize Jet and Hera machine files for section consistency
 - Fix grammar error in retrieve_data.py

 - Move date variables to J-job level
 - Keep da_obs in same directory for community and nco mode (since
process_bufr does the same)
 - In process_bufr, look for da_obs in staging directory

 - Add default directory to Jet machine file...only available there for
now
 - Add task_process_lightning to warmstart workflow

…ve to Hera. Also fix config.rrfs.yaml to better settings for test case.

…to avoid duplicate error messages, search for RRFS data on disk as well as hpss

… on Hera!

Removing the CYCLE_TYPE from config_defaults because it is a run-time,
cycle-dependent setting, not an experiment level one.

This is similar to what is done for external model files so that we can
use the machine file information about which data stores a platform has
access to.

Matching new standard outlined in Issue ufs-community#633

…obs_task

Had to manually revert some renaming changes in ush/test_retrieve_data.py

…obs_task

…obs_task

…ata isnt ready by 00z two days later

…l problems when running all unit tests

…obs_task