Fix suspicious GRIB2 table entries

[JohnHalleyGotway]

While working on issue dtcenter/MET#2780 to incorporate recent updates to GRIB2 tables into the Model Evaluation Tools software for MET version 12.0.0, I ran into the following suspicious entries in the GRIB2 tables. Recommend that you take a look.

1. NDFD_gribtable.dat has non-unique entries starting on line 27:

{0,1,0,255,8,1,19,215, "TOTALSVRPROB", "Total probability of severe thunderstorms", "??"}, 
{0,1,0,255,8,1,19,215, "TOTALXTRMPROB", "Total probability of extreme severe thunderstorms", "??"}, 

For MET, we've encoded line 28 as 216 instead of 215. However, I'm not positive that that's correct.
2. Line 108 of MRMS_gribtable.dat has the long name and units strings reversed:

{209,10,0,255,161,1,11,1, "MergedReflectivityComposite", "dBZ", "UnQCed Composite Reflectivity Mosaic (max ref)"},

Should be changed to:

{209,10,0,255,161,1,11,1, "MergedReflectivityComposite", "UnQCed Composite Reflectivity Mosaic (max ref)",  "dBZ"},

3. I also note discrepancies between the content of ncep/gribtable.dat and the GRIB2 documentation.

For example, in this table https://www.nco.ncep.noaa.gov/pmb/docs/grib2/grib2_doc/grib2_table4-2-10-2.shtml, of the entries for XICE (24), YICE (25), FRZMLTPOT (27), MLTDATE (28), and FRZDATE (29) do not appear in ncep/gribtable.dat.

@GwenChen-NOAA notes that Hera:/home/Wesley.Ebisuzaki/grib2/wgrib2/gribtables/ncep/gribtable.dat does contain entries for those. That file contains 1668 lines while ncep/gribtable.dat contains 1569. Recommend reviewing/resolving discrepancies.