Expand parallel job submissions (NOAA-EMC#28)

Expand the parallel job submissions by submitting minimization and
conventional obs plots as separate jobs using `ush/splitPlotYaml.py`.
All GFS plot jobs now to completion on hera in ~45 min using this
parallel strategy.

The `conTime.yaml` and `conVert.yaml` files have been generalized to use
DATATYPES, and use of 'all' for datatypes is now supported using the
parm/observation_datatypes.yaml master list.

The change to `ush/diagnosticAverager.py` is a pycode fix and is
unrelated to the parallel job changes.

Note that this has not been tested yet on wcoss2. I figured to do that
as a separate issue, and likely next.

Closes NOAA-EMC#25

parm/gfs/conTime.yaml

@@ -146,6 +146,8 @@ datasets:
 
     regions: 1
 
+    datatypes: {{DATATYPES}}
+
     groups:
       - name: ges
 
@@ -278,8 +280,8 @@ datasets:
 
     regions: 1
 
-    datatypes: &datatypes ['q120_0', 'q130_0']
-
+    datatypes: {{DATATYPES}}
+    
     groups:
       - name: anl
 
@@ -452,7 +454,7 @@ graphics:
     # 4 plot ConMon time series
     # -------------------------
     - batch figure:
-        datatypes: *datatypes
+        datatypes: {{DATATYPES}}
         variables:  ['count1_allev']
 
       figure:
@@ -576,7 +578,7 @@ graphics:
             label: 'final outloop'
 
     - batch figure:
-        datatypes: *datatypes
+        datatypes: {{DATATYPES}}
         variables:  ['bias1_allev']
 
       figure:

parm/gfs/conVert.yaml

@@ -24,6 +24,8 @@ datasets:
 
     regions: 1
 
+    datatypes: {{DATATYPES}}
+
     groups:
       - name: ges
 
@@ -37,7 +39,7 @@ datasets:
 
     regions: 1
 
-    datatypes: &datatypes ['q120_0', 'q130_0']
+    datatypes: {{DATATYPES}}
 
     groups:
       - name: anl
@@ -59,7 +61,7 @@ graphics:
     # 4 plot ConMon time series
     # -------------------------
     - batch figure:
-        datatypes: *datatypes
+        datatypes: {{DATATYPES}}
         variables:  ['count1', 'count2', 'count_vqc1', 'count3']
 
       figure: