Output method for UPP driver

[christinaholtNOAA]

AC

• [ ] Implementing the output method.
• [ ] Provide the output files in an appropriate data structure (dict) as a public helper method on the object.
• [ ] Update unit tests
• [ ] Add doc block to the method that shows what the output looks like (data structure with example values filled in)
• [ ] Add something to the CLI that prints out this information in a pretty-printed format (YAML). Can use it to generate docs.

Description

Use the config from the UPP to understand the naming convention of the output. Can have an "unknown" option. Returns a dictionary. Documentation of the configured output shows up in the CLI docs. Doc block that shows up in API shows an unconfigured example.

[maddenp-noaa]

Here's what we've discovered:

• The prefix for GRIB output filenames is encoded in the <datset> tag in the XML from which the postxconfig-NT.txt control file read by UPP is derived. Unfortunately, in SRW, the source XML file is not necessarily available at run time.
• But the prefix can be extracted from postxconfig-NT.txt, albeit via a somewhat byzantine method that involves reading blocks of lines and finding the required string in a specific place.
• Unfortunately, SRW and MPAS App currently use different UPP versions for which the method alluded to above must differ, given that the blocks of lines are of different sizes.
• The suffix for GRIB output filenames appears to be decided in Fortran code based on a difficult-to-decipher recipe, but for the applications we're currently interested in, appears to be .GrbF*, where the * is the zero-padded forecast leadtime of at least 2 digits (e.g. .GrbF06, .GrbF168, etc.).
• We opened a discussion in the UPP repo to inquire about any straightforward way to predict the output filenames based on the input configuration.

Current plan of action:

• Implement the prefix-extraction method assuming the longer line blocks of the sort associated with the newer UPP version used by MPAS App.
• Generate suffixes based on what is currently observed from SRW and MPAS App, although we are aware this is probably not sufficiently general.
• Update the uwtools UPP driver to make the path to the postxconfig-NT.txt control file available under a required and schema-checked key, so that it can be consulted by the output() method.
• Implement output() as a function of postxconfig-NT.txt and the forecast leadtime, known to the driver.