em_convrad case has time-varying solar radiation

[tschanzer]

Issue
Output from the em_convrad case shows a diurnal cycle of incoming solar radiation, contrary to the definition of RCE (e.g., GMD - Radiative–convective equilibrium model intercomparison project). Is this intended behaviour?

Steps to reproduce

1. Download WRF v4.6.0 source code (specifically, the version ported for Australian supercomputer Gadi -- GitHub - coecms/WRF at V4.6.0)
2. Configure for serial mode and no nesting
3. Compile for em_convrad case
4. Change test/em_convrad/namelist.input so the run length is 1 day and domain size is 11x11, leaving other settings the same (see attached)
5. Run test/em_convrad/ideal.exe
6. Run test/em_convrad/wrf.exe
7. Inspect time series of domain averages of output variables SWDNT (downwelling shortwave flux at top) and COSZEN (cosine of solar zenith angle) using Python and xarray (figures attached):

   import xarray as xr
   ds = xr.open_dataset("<wrf dir>/test/em_convrad/wrfout_d01_2007-06-01_00:00:00")
   ds["SWDNT"].mean(["south_north", "west_east"]).plot()
   ds["COSZEN"].mean(["south_north", "west_east"]).plot()