Some changes needed for NUOPC cap (version in CESM2.2.0)

Some more changes from @mvertens with updates for the NUOPC cap.
There's a new nuopc_cap_share directory added. SetVM is made public
in rof_comp_nuopc.F90 and some of the logging is slightly changed.
RtmIO is now using ROFID rather than instance name.

Also convert the nldas2 file into NetCDF5 format from Jim Edwards.

Bring in mosart1_0_35 to release branch -- cold start fix, few other small changes

Fix for cold start and add cold start test

Change wallclock time in tests and change hobart for izumi

Add code of conduct

Change all instances of SHR_KIND_CL to use CL that is renamed from
shr_kind_cl in shr_kind_mod. This is important for a recent cime
update.

Remove ability to write out single-precision history because of issues

Don't allow the namelist option rtmhist_ndens to be set to 2, because this
option doesn't currently function. The simple fix we put into place for it
is not robust.

Fix for py3

Fix for python3, using the floor operator for integer division.

New r8th routing file, pylint and py3 checking, fill value needed for pio2

Add in 8th degree routine file. Run pylint and check for py3 compatability.
There's explict setting of fill type. And also explicit use of shr_kind_r4 for
kind rather than real(4), which is a better mechanism. Also ncd_getiodesc
will read in PIO_DOUBLE for input xtype= PIO_DOUBLE or PIO_REAL. Most of this
is direct from jedwards4b (other than r8th addition).

First CESM2.0 release tag, identical to mosart1_0_31

Science changes since: mosart1_0_00

• Changes from Sean Swenson to add minimum value to rlen
• Update areas on routing file
• Answer changing improvements to channel storage from HongYi Li for faster spinup.
• Bugfix from Tony Craig to use correct delta time for qgwl flux.
• Treat irrigation as a seperate flux
• bugfix for budget diagnostic output
• bugfix: calculation of qgwl_volume must be multiplied by ar
• Update default routing file.
• New method of handeling runoff terms to avoid negative runoff.
• switch to input dataset MOSART_Global_half_20151130a.nc
• update direct sparse matrix to include non basin points in order
  to pass data from any grid cell directly to the ocean.
• modify the direct term and push all direct water to outlet points
• set all tracer 2 water (frozen water) to be a direct term
• add ability to skip some tracers in the euler solver via euler_calc flag
• add a budget accumulator term
• Fix exact restart in atan slope calc
• Swenson bugfix for mosart direction file
• Swenson river volume normalization bugf

Software changes since: mosart1_0_00

• Add config_archive for mosart
• delete rof_comp_esmf
• add model_doi_url,
• change some instances of RTM/CLM in documentation to MOSART
• Update testlist to version 2 format, remove ys tests
• Upgrade config_component to version 3
• allow output file format to change
• If NINST_ROF > 1, check if instance number in name for REFCASE
• Update testlist and use Clm5 compset names
• Update routing file
• Update to cime5 python namelist infrastructure
• Fix an issue with nag, lower amount of log output
• Add output frequency to history files
• Turn off for CLM_ACCELERATED_SPINUP="on"
• Add new namelist options bypass_routing_option,
  qgwl_runoff_option
• Error checking on max length of history filenames.
• Add testdefs dir, testmods and testlist for integration to cime
  test system.
• remove rofdto from coupler interface fields.
• Changes to get Mosart to build with Nag on Hobart
• merge ACME fixes to decomp and performance
• cime compatible infrastructure
• Add direct to ocean runoff flux
• PIO2 updates