Streamlining about g-point-reduction HOT 2 OPEN

regarding the template used as input into the driver, here is an abridged version of my current protocol:

1. grab separate single-band and broadband templates in /project/projectdirs/e3sm/pernak18/reference_netCDF/g-point-reduce/
2. input single-band file from item 1 in the Garand RRTMGP driver in this repository (working version from Dec 2020, not the most recent commit in garand_atmos)
3. run with broadband file from 1 in the same executable as 2
4. generate a shortwave template, which has other input parameters in the netCDF
5. repeat steps 2 and 3 for the SW

this is a bit tedious somewhat difficult to remember, plus it hinders our examination with the profz and statz plotz, because our output flux netCDFs from steps 3 and 5 are only correct in the broadband (by-band fluxes could conceivably be correct as well, but then broadband fluxes would be wrong).

i'm working on a workaround that will stitch together correct by-band results from 2 with correct broadband results from 3 (and will probably do the same thing for the SW) so we have a central reference file with both, which we do not have right now.

from g-point-reduction.

more commentary on the template in another issue:

from g-point-reduction.