fix float invalidation error of eca_bgc

[jinyuntang]

By initializing col_plant_pdmeand_vr properly, it fixes
a NaN triggered error of float invalidation when running
ECACNP with DEBUG=TRUE.

[non-BFB]
Fixes #3296

[dqwu]

@bishtgautam
This PR has caused two tests to fail on melvin:

ERS.f19_g16.I1850GSWCNPECACNTBC.melvin_gnu.clm-eca_f19_g16_I1850GSWCNPECACNTBC
ERS.f19_g16.I20TRGSWCNPECACNTBC.melvin_gnu.clm-eca_f19_g16_I20TRGSWCNPECACNTBC

See https://my.cdash.org/viewTest.php?onlydelta&buildid=1726125

Looks like baseline mismatch:

  comparing model 'cpl'
    /ascldap/users/jenkins/acme/scratch/ERS.f19_g16.I1850GSWCNPECACNTBC.melvin_gnu.clm-eca_f19_g16_I1850GSWCNPECACNTBC.C.JNextDeveloper20191108_221231/run/ERS.f19_g16.I1850GSWCNPECACNTBC.melvin_gnu.clm-eca_f19_g16_I1850GSWCNPECACNTBC.C.JNextDeveloper20191108_221231.cpl.hi.0001-01-12-00000.nc.base did NOT match /ascldap/users/jenkins/acme/scratch/ERS.f19_g16.I1850GSWCNPECACNTBC.melvin_gnu.clm-eca_f19_g16_I1850GSWCNPECACNTBC.C.JNextDeveloper20191108_221231/run/ERS.f19_g16.I1850GSWCNPECACNTBC.melvin_gnu.clm-eca_f19_g16_I1850GSWCNPECACNTBC.C.JNextDeveloper20191108_221231.cpl.hi.0001-01-12-00000.nc.rest
    cat /ascldap/users/jenkins/acme/scratch/ERS.f19_g16.I1850GSWCNPECACNTBC.melvin_gnu.clm-eca_f19_g16_I1850GSWCNPECACNTBC.C.JNextDeveloper20191108_221231/run/ERS.f19_g16.I1850GSWCNPECACNTBC.melvin_gnu.clm-eca_f19_g16_I1850GSWCNPECACNTBC.C.JNextDeveloper20191108_221231.cpl.hi.0001-01-12-00000.nc.base.cprnc.out
FAIL
...
  comparing model 'cpl'
    /ascldap/users/jenkins/acme/scratch/ERS.f19_g16.I1850GSWCNPECACNTBC.melvin_gnu.clm-eca_f19_g16_I1850GSWCNPECACNTBC.C.JNextDeveloper20191108_221231/run/ERS.f19_g16.I1850GSWCNPECACNTBC.melvin_gnu.clm-eca_f19_g16_I1850GSWCNPECACNTBC.C.JNextDeveloper20191108_221231.cpl.hi.0001-01-12-00000.nc did NOT match /sems-data-store/ACME/baselines/gnu/master/ERS.f19_g16.I1850GSWCNPECACNTBC.melvin_gnu.clm-eca_f19_g16_I1850GSWCNPECACNTBC/cpl.hi.0001-01-12-00000.nc
    cat /ascldap/users/jenkins/acme/scratch/ERS.f19_g16.I1850GSWCNPECACNTBC.melvin_gnu.clm-eca_f19_g16_I1850GSWCNPECACNTBC.C.JNextDeveloper20191108_221231/run/ERS.f19_g16.I1850GSWCNPECACNTBC.melvin_gnu.clm-eca_f19_g16_I1850GSWCNPECACNTBC.C.JNextDeveloper20191108_221231.cpl.hi.0001-01-12-00000.nc.cprnc.out
FAIL

[jinyuntang]

@dqwu and @bishtgautam, I suspect depending on the platform, the baseline may need to be regenerated. In the default code (i.e., the master), the array col_plant_pdemand_vr is not initialized properly, and depending on the compiler and platform, the value of col_plant_pdemand_vr at the 1st time step is totally random.

[bishtgautam]

Ok, I'm going to change this PR to be non-BFB.

[bishtgautam]

@jinyuntang This PR broke the following ECA ERS tests:

• ERS.f19_g16.I1850GSWCNPECACNTBC.melvin_gnu.clm-eca_f19_g16_I1850GSWCNPECACNTBC
• ERS.f19_g16.I20TRGSWCNPECACNTBC.melvin_gnu.clm-eca_f19_g16_I20TRGSWCNPECACNTBC

See https://my.cdash.org/viewTest.php?onlyfailed&buildid=1727482. I have reverted this PR.

[jinyuntang]

@bishtgautam and @rljacob , I saw the following line in the reports of both simulations.
Error: Computation time increase > 10 pct from baseline
How do we handle this usually? I assume I will leave this alone at the moment.

[rljacob]

It worth looking at the timers more closely.

[rljacob]

@jinyuntang did you look at the timers?

[jinyuntang]

@rljacob, I did not get a chance to do this. But I have asked @qzhu-lbl to bundle it with his v2.ECA branch, because it also affects that branch.

[rljacob]

So this PR should be closed because the work will be done in a different, future PR?

[jinyuntang]

@rljacob, I think so.

[jinyuntang]

@rljacob, but please wait till @qzhu-lbl makes this change in the v2.ECA code.

[rljacob]

Closing this won't affect the other branch. We can leave the branch intact.