* Fix for duplicate variables in xarray DataTree

Added logic to make sure that GRIB2 metadata be of type int, not
Grib2Message when adding these as separate columns to pandas df
for datatree processing

Added entries for 195, 196, 197 in level table for datatree. These
are custom levels used by AWC (and thus NBM).

[skip ci]

* Significant Update for xarray_backend.py

This commit allows for the backend to provide a common dimension name
for GRIB2 messages where more than 1 attribute would provide the dimensionality.

For example, probability messages (pdtn 5 or 9), these messages have the following
attrs used as dimensions: thresholdLowerLimit, thresholdUpperLimit, but these
are valid for 1 message, basically we are adding an extra dimension.

So here we introduce a "threshold" dimension (and coordinate), and thresholdLowerLimit
and thresholdUpperLimit serve as coordinates, dimensioned by "threshold".

Also, this commit adds the ability to add another node for Xarray DataTrees by the
typeOfProbability attribute, providing node name "prob_<typeOfProbability>".

* Update xarray_backend.py

Fixed bug from previous commit where for mfdatasets the dimension ordering
was wrong when creating the DataArray.

The fix is reversing ordered_meta in make_variables().

* Update xarray_backend.py

Cleaned up *Dim class naming for DimCube.

* Update xarray_backend.py

For *Dim classes, using typing.Tuple[] instead of tuple[].  This maintains
compatibility with Python 3.8.

* Update for xarray_backend.py

This commit comments out the call to create_dataset_from_df() and instead
calls try_process_by_variables().

This leads to variables for a level/pdtn tree gettings their own var_<shortName>
DataTree group.

The next step here is to collect DataArrays where they have the same level value.
For example, collect temps where 2m height above ground and winds where 10m
height above ground.

With these changes, all GRIB2 messages are resolved when reading a NBM Core F001
CONUS GRIB2 file.

The multitude of diagnostic print statements are still present.

* additions to pr:200 duplicate elements with open datatree (#203)

* this appears to be a working checkpoint

* revisions and cleanup for adding extra coordinates for dimensions that are not indexes; level, threshold

---------

Co-authored-by: Adam.Schnapp <[email protected]>

* Update templates.py

Added check in __get__ for valueOfFirstFixedSurface and valueOfSecondFixedSurface
to check for scale_factor < 0, to return 0.0

[skip ci]

* 200 duplicate elements with open datatree (#204)

* this appears to be a working checkpoint

* revisions and cleanup for adding extra coordinates for dimensions that are not indexes; level, threshold

* Update templates.py

Added check in __get__ for valueOfFirstFixedSurface and valueOfSecondFixedSurface
to check for scale_factor < 0, to return 0.0

[skip ci]

* adjust a test for adjusted dimension/coordinate behavior and fix bug parsing grib index

---------

Co-authored-by: Adam.Schnapp <[email protected]>
Co-authored-by: Eric Engle <[email protected]>

* Add Additional Metadata and Data Model Conversions (#201)

* Add support for additional coordinate and metadata information
in xarray dataset. Add optional data_model argument
to translate coordinates and attributes to
defined data model specification.

* integrate aspects of the cf_metadata into the default data model

* Move ptype threshold decoding to data model function

* Fix string type conversion in metadata parsing

* Add comment with reference to CF conventions standard names table

* Fix error in ptype parsing and remove forced conversion to string for metadata values

* Add else to rename all coordinate names

* Change leadtime to lead_time per
[email protected]

---------

Co-authored-by: Adam.Schnapp <[email protected]>

* Updating grib2io tables

Added reworked CF tables

NCEP GRIB2 tables updated to v35

[skip ci]

* Updates to xarray backend.

Moved table in vertical_coordinate_surfaces.py to inside
xarray_backend.py.

Code cleanup.

[skip ci]

* Update for tables and table generation scripts.

[skip ci]

* Update GRIB2 tables

Updated make_grib2_tables/get-ncep-grib2-originating-centers.py to
perform more explicit table scraping and encoding due to pandas
read_html not properly encoding characters with accent marks.

[skip ci]

* Update for xarray_backend.py

Changes here for the Xarray DataTree.  As of this commit, all GRIB2
messages for a Core NBM GRIB2 files are accounted for.

[skip ci]

* Update xarray_backend.py

Clean up of DataTree names.

[skip ci]

* Update tests due to NCEP GRIB2 table updates.

[skip ci]

* Update tests/test_xarray_datatree_backend.py

Xarray DataTree .ds is not None ever, but will return a DatasetView
of an empty Dataset when there is none.

* Clean up xarray_backend.py

Remove test prints.

[skip ci]

* More datatree updates

[skip ci]

* Update xarray_backend.py

[skip ci]

* Update tests for xarray DataTree backend

Added file blend.t00z.core.f001.co_4x_reduce.grib2 which is a NBM
Core forecast GRIB2 file for CONUS where the 2.5km grid has been
rediced 4x to roughly 10km to make the filesize smaller, but all GRIB2
metadata are preserved.

Added test to resolve all messages in the file.

* Update DataTree tests

Commented out some check due to changes to the DataTree structure.

[skip ci]

* Update tests.

---------

Co-authored-by: Eric Engle <[email protected]>
Co-authored-by: Adam Schnapp <[email protected]>
Co-authored-by: Adam.Schnapp <[email protected]>
Co-authored-by: TylerWixtrom-NOAA <[email protected]>