Add lsm_cold_start flag (RUC LSM) and gwdps bugfix (combined)

[grantfirl]

This PR adds a lsm_cold_start variable that is TRUE when the LSM is supposed to be cold-started. This replaces the use of (.not. flag_for_restart) in the RUC LSM code and GFS_phys_time_vary_init subroutine.

Fixes #833

[tanyasmirnova]

@grantfirl Grant, the introduction of lsm_cold_start instead of restart make a lot of sense to me. It uncouples atmosphere and soil, for example, if we cold-start FV3's atmosphere, we could set lsm_cold_start=.false., thus the soil variables will be continuously cycling. We need this capability in the cycled RRFS. I think the follow-up change could be done also in FV3GFS_io.F90 to use lsm_cold_start as a control for the input of LSM variables.

[SamuelTrahanNOAA]

For backward-compatibility, the default value for lsm_cold_start needs to be .not.flag_restart or existing workflows will fail.

[SamuelTrahanNOAA]

Expecting people to update their workflows is okay, I guess, but we should avoid it if possible.

[grantfirl]

For backward-compatibility, the default value for lsm_cold_start needs to be .not.flag_restart or existing workflows will fail.

Agreed. I set the default values in both the SCM and FV3 to .not. flag_restart. See, e.g. https://github.com/NCAR/ccpp-scm/pull/297/files#diff-5164aebefa34d8f4d2bb13dce168e12fd89d4a2fcf4badf6b9fa720fb0c2fe4e and https://github.com/NOAA-EMC/fv3atm/pull/475/files#diff-89fcd58cb0c9d86e79683359371e2dc13cefa09bc447a65e01e58ba591d89269

[grantfirl]

Combined #832 into this PR. Fixes #831

[grantfirl]

@SamuelTrahanNOAA Since you reviewed this, could you please approve? I've combined this with #832 for UFS testing and the UFS code managers are working on this today.