Comments about run-off fluxes are overly specific to JRA55

[glwagner]

There are comments in the code that are overly specific to JRA55 reanalysis. Given that we will eventually support many different prescribed atmospheres, including both JRA55 and ERA5, which presumably have a variety of ways of prescribing run-off, etc, this code needs to be generalized somehow.

ClimaOcean.jl/src/OceanSeaIceModels/CrossRealmFluxes/atmosphere_ocean_fluxes.jl

Lines 341 to 368 in 10a3375

	# Retrieve the details of runoff fluxes (rivers and icebergs, if present in the simulation).
	# Note that these forcing fields are different in terms of frequency (daily instead of three-hourly)
	# and gridsize (1/4 degree instead of 1/2 degree) when compared to the other prescribed fluxes
	# So they need to be interpolated using their own grid / times / backend / time_indexing
	@inline get_runoff_args(::Nothing) = nothing
	@inline function get_runoff_args(runoff_flux)
	data = map(ϕ -> ϕ.data, runoff_flux)
	grid = runoff_flux.rivers.grid
	times = runoff_flux.rivers.times
	backend = runoff_flux.rivers.backend
	time_indexing = runoff_flux.rivers.time_indexing
	return (data, grid, times, backend, time_indexing)
	end
	@inline get_runoff_flux(X, time, ::Nothing) = zero(eltype(X))
	@inline function get_runoff_flux(X, time, runoff_args)
	@inbounds runoff_flux = runoff_args[1] # The data is located at position 1 of the tuple
	@inbounds other_args = runoff_args[2:end] # Other args contain grid, times, backend and time_indexing
	Mr = interp_atmos_time_series(runoff_flux, X, time, other_args...)
	return Mr
	end

[glwagner]

This comment also should be removed:

ClimaOcean.jl/src/OceanSeaIceModels/CrossRealmFluxes/atmosphere_ocean_fluxes.jl

Line 261 in 10a3375

# Rememeber runoff fluxes could be `nothing` if rivers and icebergs are not included in the forcing

because any of the fluxes can be nothing --- why not?

[glwagner]

Here are two solutions:

1. Every variable has it's own grid.
2. Every variable lives on the same spatial grid. This requires interpolating the atmos variables to a common grid when constucting them.

It's too complicated to pick a random subset of variables (like runoff) to have a different grid, this doesn't generalize to many different reanalysis products.