FAIL SMS_D_Ld3_PS.f09_t233.I1850Clm60BgcNoAnthroRtm.derecho_intel.clm-decStart1851_noinitial--clm-matrixcnOn #4263

This issue documents a failure that may affect Paleo users (most likely), when running with matrixcnOn and the compset of this aux_clm test:
SMS_D_Ld3_PS.f09_t233.I1850Clm60BgcNoAnthroRtm.derecho_intel.clm-decStart1851_noinitial--clm-matrixcnOn.

• The test failed consistently in PR ctsm5.4.067: Move CTSM tests from gx grids to t233 grid #4234 with "floating invalid" in line 764 CNSoilMatrixMod.F90, not line 793 of the tests that pass by resubmitting a few times
• The test's predecessor with f09_g17 passed in the baseline (ctsm5.4.063) and again while testing PR ctsm5.4.067: Move CTSM tests from gx grids to t233 grid #4234
• In RTM's testlist we use f09_f09_mt233 ONLY (instead of f09_t233), but the test fails also with that
• In RTM's testlist we never test BgcNoAnthroRtm (so only SpNoAnthroRtm, so we didn't catch the problem in the corresponding RTM PR); that's an omission, though ok since we test BgcNoAnthroRtm in aux_clm
• Testing RTM with matrixcnOn is unnecessary because we use the latter ONLY in SASU mode (a phase of the BGC accelerated spin-up), during which RTM preferably remains unset to conserve compute cycles
• The test without matrixcnOn passes, so I updated this aux_clm test by removing matrixcnOn

General bug information

CTSM version you are using: PR #4234 testing against ctsm5.4.063 baseline

Does this bug cause significantly incorrect results in the model's science? No

Configurations affected: f09_t233.I1850Clm60BgcNoAnthroRtm with matrixcnOn