`ECCOPrescribedAtmosphere` does not support `dir` argument

[xkykai]

the ECCOPrescibedAtmosphere API currently does not allow users to specify the dir kw argument which stipulates where the ECCO atmospheric fields should be stored. Here's an MWE:

using ClimaOcean
using ClimaOcean.DataWrangling.ECCO
using Oceananigans
using Dates

arch = CPU()
dir = joinpath(homedir(), "forcing_data")
start_date = DateTime(1992)
end_date = start_date + Month(3)
time_indices_in_memory = 3

atmosphere = ECCOPrescribedAtmosphere(arch; dir, start_date, end_date, time_indices_in_memory)

which throws

ERROR: MethodError: no method matching FieldTimeSeries(::Metadata{…}, ::LatitudeLongitudeGrid{…}; time_indices_in_memory::Int64, time_indexing::Oceananigans.OutputReaders.Cyclical{…}, dir::String)
This method does not support all of the given keyword arguments (and may not support any).

Closest candidates are:
  FieldTimeSeries(::Metadata, ::Oceananigans.Grids.AbstractGrid; time_indices_in_memory, time_indexing, inpainting, cache_inpainted_data) got unsupported keyword argument "dir"
   @ ClimaOcean ~/.julia/packages/ClimaOcean/tO1vi/src/DataWrangling/metadata_field_time_series.jl:109
  FieldTimeSeries(::String, ::Any...; reader_kw, kwargs...)
   @ Oceananigans ~/.julia/packages/Oceananigans/t5Rdl/src/OutputReaders/field_time_series.jl:869
  FieldTimeSeries(::Metadata; ...)
   @ ClimaOcean ~/.julia/packages/ClimaOcean/tO1vi/src/DataWrangling/metadata_field_time_series.jl:103
  ...

Stacktrace:
 [1] kwerr(::@NamedTuple{…}, ::Type, ::Metadata{…}, ::LatitudeLongitudeGrid{…})
   @ Base ./error.jl:175
 [2] FieldTimeSeries(metadata::Metadata{…}, arch::CPU; kw::@Kwargs{…})
   @ ClimaOcean.DataWrangling ~/.julia/packages/ClimaOcean/tO1vi/src/DataWrangling/metadata_field_time_series.jl:106
 [3] ECCOPrescribedAtmosphere(architecture::CPU, FT::Type; dataset::ECCO4Monthly, start_date::DateTime, end_date::DateTime, time_indexing::Oceananigans.OutputReaders.Cyclical{…}, time_indices_in_memory::Int64, surface_layer_height::Int64, other_kw::@Kwargs{…})
   @ ClimaOcean.DataWrangling.ECCO ~/.julia/packages/ClimaOcean/tO1vi/src/DataWrangling/ECCO/ECCO_atmosphere.jl:44
 [4] ECCOPrescribedAtmosphere
   @ ~/.julia/packages/ClimaOcean/tO1vi/src/DataWrangling/ECCO/ECCO_atmosphere.jl:23 [inlined]
 [5] top-level scope
   @ REPL[21]:1
Some type information was truncated. Use `show(err)` to see complete types.

[glwagner]

When should we move development to NumericalEarth? cc @simone-silvestri

[simone-silvestri]

Yeah I think we should move it now given all these bugfixes that will be lost when we connect the repos.
@xkykai would it be too much of a hassle to use NumericalEarth instead of ClimaOcean while we figure out how to connect the two repos in a proper, sustainable way? The contents of the repo as of now should be the same so hopefully it should not require too much work to switch from one to another.