TODO

FEATURES
 - Option to have more McICA columns to reduce noise
 - Compute UV index when sufficient UV bands

LONGER TERM FEATURES
 - Option to provide date, time, longitude and latitude from which the
   IFS values will be calculated of: solar zenith angle, trace gas and
   aerosol mixing ratio, and cloud effective ratio
 - Incorporate RRTMGP
 - Solar cycle in spectral irradiance

INTERFACE
 - Rename "do_3d" as it is misleading
 - Add deallocate method for config type
 - Column is last array argument for all input and output arrays?
 - Why are the flux%*_band variables dimensioned (g,col,lev) not (g,lev,col)?

ECCKD AND GENERALIZED CLOUD/AEROSOL

DOCUMENTATION
 - Add chapters on API, internal structure and underpinning science

MISC
 - Definition of Tdir in two_stream

DIFFERENCES IN IFS VERSION OF ECRAD
 - Several files contain easy_netcdf_with_read_mpi instead of
   easy_netcdf (radiation_aerosol_optics_data,
   radiation_cloud_optics_data, radiation_pdf_sampler)
 - radiation_cloud_generator.F90 uses SIMD routine to sample PDF
 - radiation_delta_eddington.h contains SIMD routine
 - radiation_interface.F90 and radiation_ifs_rrtm.F90 take YDERAD
   arguments, the latter has other differences
 - radiation_constants should use IFS values?

KNOWN BUGS