Ocean/global_stats: Removed land cells from stats, added readme file

[mark-petersen]

The global_stats branch is now ready for review. Alterations include:

1. Addition of stats_readme.txt file
2. remove land cells from computations
3. remove white space from stats_time.txt

I tested it with the 120km global and 10km overflow. Both match output files bit-for-bit with master.

Min and Avg stats are no longer contaminated by -1e34:
120km, master branch. Columns are temperature, salinity, tracer1

lo1-fe.lanl.gov> cut -c 363-500 stats_max.txt | tail -n 1
2.86914922866561E+01 3.74667462059278E+01 1.00000000000000E+00
lo1-fe.lanl.gov> cut -c 363-500 stats_avg.txt | tail -n 1
-3.08606381137871E+33 -3.08606381137871E+33 -3.08606381137871E+33
lo1-fe.lanl.gov> cut -c 363-500 stats_min.txt | tail -n 1
-1.00000000000000E+34 -1.00000000000000E+34 -1.00000000000000E+34

120km, global_stats branch

lo1-fe.lanl.gov> cut -c 363-500 stats_max.txt | tail -n 1
2.86914922866561E+01 3.74667462059278E+01 1.00000000000000E+00
lo1-fe.lanl.gov> cut -c 363-500 stats_avg.txt | tail -n 1
3.40931804591106E+00 3.47108117173064E+01 1.00000000000000E+00
lo1-fe.lanl.gov> cut -c 363-500 stats_min.txt | tail -n 1
-1.74337564684972E+00 2.45582429725827E+01 9.99999999999998E-01

[douglasjacobsen]

I tested, and the results are bit-for-bit identical to the MPAS-Dev/develop.

With one exception.

The changes to mpas_ocn_mpas_core.F in commit d4c80b4 change the initial output frame. These changes cause tracer values on land cells (in a column with some ocean cells) to have an initial value of -1e34, so differences with develop in the initial frame are of the order 1e33 in tracers.

This also fixes item 4 on our task lists related to the first output frame not containing -1e34 like later frames do.

I used a 240km ocean grid to test.