cprnc: allow differences in field lists for time-constant fields

[billsacks]

In cprnc: For files with an unlimited (time) dimension: Separately count
(1) missing time-varying fields and (2) missing time-constant
fields. Only (1) is considered in determining whether to report a final
difference in the field lists.

Before this, no distinction was made between time-varying
vs. time-constant fields in counting the number of missing
variables. (These counts were added in #2988). However, that led to
failures in some exact restart tests, because some time-constant fields
were on output files from one case but not the other (see #3007).

Here is sample output from cprnc for a few cases:

(1) Difference in the presence / absence of time-varying variables:

SUMMARY of cprnc:
 A total number of      6 fields were compared
          of which      0 had non-zero differences
               and      0 had differences in fill patterns
               and      0 had different dimension sizes
 A total number of      0 fields could not be analyzed
 A total number of      2 time-varying fields on file 1 were not found on file 2.
 A total number of      0 time-constant fields on file 1 were not found on file 2.
 A total number of      0 time-varying fields on file 2 were not found on file 1.
 A total number of      0 time-constant fields on file 2 were not found on file 1.
  diff_test: the two files DIFFER only in their field lists

(2) Difference in the presence / absence of time-constant variables, for
files that have a time dimension:

SUMMARY of cprnc:
 A total number of     13 fields were compared
          of which      0 had non-zero differences
               and      0 had differences in fill patterns
               and      0 had different dimension sizes
 A total number of      0 fields could not be analyzed
 A total number of      0 time-varying fields on file 1 were not found on file 2.
 A total number of      2 time-constant fields on file 1 were not found on file 2.
 A total number of      0 time-varying fields on file 2 were not found on file 1.
 A total number of      1 time-constant fields on file 2 were not found on file 1.
  diff_test: the two files seem to be IDENTICAL
      (But note that there were differences in field lists just for time-constant fields.)

(3) Difference in the presence / absence of time-constant variables, for
files that do NOT have a time dimension (note that this still results in
a DIFFER result; it seemed to me that that's what a user would want in
this case):

SUMMARY of cprnc:
 A total number of      5 fields were compared
          of which      0 had non-zero differences
               and      0 had differences in fill patterns
               and      0 had different dimension sizes
 A total number of      0 fields could not be analyzed
 A total number of      2 fields on file 1 were not found on file 2.
 A total number of      1 fields on file 2 were not found on file 1.
  diff_test: the two files DIFFER only in their field lists

The downside of this solution is that we wouldn't catch removals of
time-constant fields when doing baseline comparisons, and might overlook
this for interactive uses of cprnc. To address the issue of baseline
comparisons, I thought about adding a flag that treats time-constant
fieldlist diffs as differences
(#3007 (comment)), but
I have not done this here because this would have taken more time to
implement and added significantly more complexity (in both the cprnc
Fortran and in hist_utils.py) for uncertain benefit. (See the discussion in #3007
for more thoughts on this.)

Once this PR is approved and merged, I will update the cprnc builds on
hobart and cheyenne (currently they are using builds prior to #2988).

Test suite:

• Ran the two tests noted as failing in ERS and ERP tests failing with updated cprnc #3007, pointing to a version of
  cprnc built from this branch; these tests now pass.
• scripts_regression_tests on cheyenne

Test baseline: n/a
Test namelist changes: none
Test status: bit for bit

Fixes #3007

User interface changes?: Changes cprnc behavior slightly, as noted above

Update gh-pages html (Y/N)?: N

Code review: @jedwards4b @jgfouca @fischer-ncar

Also cc @rljacob @gold2718 @mnlevy1981 @ekluzek

[billsacks]

scripts_regression_tests.py has completed on cheyenne. The only failures are ones that also failed in the baseline (cime5.7.9, on which this branch is based):

======================================================================
FAIL: test_appends_not_overriden (__main__.H_TestMakeMacros)
Test that machine-specific base value changes don't interfere with appends.
----------------------------------------------------------------------
Traceback (most recent call last):
  File "./scripts_regression_tests.py", line 3074, in test_appends_not_overriden
    tester.assert_variable_equals("FFLAGS", "-base2", env={"COMPILER": self.test_compiler})
  File "./scripts_regression_tests.py", line 2566, in assert_variable_equals
    self.parent.assertEqual(self.query_var(var_name, env, var), value)
AssertionError: '-base2 -debug1' != '-base2'

======================================================================
FAIL: test_appends_not_overriden (__main__.I_TestCMakeMacros)
Test that machine-specific base value changes don't interfere with appends.
----------------------------------------------------------------------
Traceback (most recent call last):
  File "./scripts_regression_tests.py", line 3074, in test_appends_not_overriden
    tester.assert_variable_equals("FFLAGS", "-base2", env={"COMPILER": self.test_compiler})
  File "./scripts_regression_tests.py", line 2661, in assert_variable_equals
    self.parent.assertEqual(self.query_var(var_name, env, var), value)
AssertionError: '-base2 -debug1' != '-base2'

----------------------------------------------------------------------
Ran 149 tests in 3281.197s

FAILED (failures=2, skipped=10)

[jedwards4b]

The tests you mention as failing both pass for me on master.

[billsacks]

@jedwards4b - @fischer-ncar removed himself as a reviewer. Given that you have already reviewed this, I have reassigned this to you.