do not report memcomp if no baseline exists

[jedwards4b]

This PR changes the test error:
FAIL SMS_D_Ld3_PS.f09_g17.I1850Clm60SpNoAnthro.derecho_intel.clm-decStart1851_noinitial--clm-nofireemis MEMCOMP [Errno 2] No such file or directory: '/glade/campaign/cesm/cesmdata/cesm_baselines/cesm3_0_beta03/SMS_D_Ld3_PS.f09_g17.I1850Clm60SpNoAnthro.derecho_intel.clm-decStart1851_noinitial--clm-nofireemis/cpl-mem.log' and does not report
MEMCOMP if no baseline cpl-mem.log exists.

Test suite: scripts_regression_tests.py, cesm prealpha
Test baseline:
Test namelist changes:
Test status: bit for bit

Fixes #4590

User interface changes?: N

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

[jedwards4b]

@jasonb5 I'm not sure if it would be better to somehow report MEMCOMP but not fail and not pass or just not report if there is no baseline file present. I chose not to report just because that's easier.

[jasonb5]

@jedwards4b I think it's probably fine to not report if there is no baseline, just need to test the behavior.

[jasonb5]

@rljacob @jgfouca I tested this and if the the baseline files are missing then the phase (MEMCOMP, TPUTCOMP) is omitted from the TestStatus file. If the memory baseline is missing then the MEMCOMP is omitted but the MEMLEAK is still included. I'm wondering if as an alternative to omitting it might be better to introduce a new status of IGNORE or SKIP when the baseline is missing? Or are we ok with the phases being omitted?

[rljacob]

omitted is fine with me.

[jasonb5]

LGTM