negative C/N error

While using CLM50%BGC-CROP for an FHIST run on ne30_ne30 grid, the model crashes (within ~few days-1 month) with a negative C/N error. The default time step for the ne30 grid is 1800s (ATM_NCPL=48). By reducing the time step to 600s (ATM_NCPL=144), the error goes away and the model can run ~5 years successfully.

After discussion with @dlawrenncar, we tried using larger C/N negative error bounds as well as disabling the C/N negative error check in CNPrecisionControlMod. Simulations without error check and a coarse time step (ATM_NCPL=48) fails after ~9 months with a water balance error.

### General bug information

**CTSM version you are using:** CLM50%BGC-CROP in cesm2.0.0

**Does this bug cause significantly incorrect results in the model's science?** Yes

**Configurations affected:** ne30_ne30_mg16 grid

### Important details of your setup / configuration so we can reproduce the bug
compset - FHIST
resolution - mg16_me16_ne30
machine - NERSC Cori, run using 64 nodes (2048 proc.)
./xmlchange SSTICE_DATA_FILENAME=/project/projectdirs/ccsm1/inputdata/atm/cam/sst/sst_HadOIBl_bc_1x1_1850_2016_c170525.nc

user_nl_clm:
use_init_interp = .true.
init_interp_fill_missing_with_natveg = .true.
create_crop_landunit = .true.
check_finidat_year_consistency = .false.

flanduse_timeseries = ""
fsurdat = "surfdata_ne30np4_78pfts_CMIP6_simyr1850_c170824.nc"
finidat='clm50_r270_1deg_GSWP3V1_iso_newpopd_hist.clm2.r.2015-01-01-00000.nc'

### Important output or errors that show the problem
Error message in cesm log says
ERROR: Carbon or Nitrogen patch negative = -60.1641313653330 -1.20755484588496 ERROR: limits = -60.0000000000000 -6.00000000000000