Custom Struct as outtype in OutDims?

Start with YAXArrays.DAT registration.jl at the OutDims constructors shown in the stack trace, then compare the successful foo example with MaximumLikelihoodAbstractExtremeValueModel. Reproduce both OutDims calls and inspect the conversion failure. Done means determining whether the custom model type can be accepted consistently as outtype and documenting or testing the resulting behavior.

Hello,

I am trying to do GEV estimation using mapCube, using the following logic. The objective is to get the timeseries for each gridpoints and return the GEV model associated with the timeserie, for each gridpoint, storing the following information into a YAXArray:

MaximumLikelihoodAbstractExtremeValueModel
model :
        ThresholdExceedance
        data :          Vector{Float64}[28]
        logscale :      ϕ ~ 1
        shape :         ξ ~ 1

θ̂  :    [2.0154230098844907, 0.09042541506171095]

Hence, when I try the following:

indims = InDims("Ti")
outdims = OutDims(outtype=MaximumLikelihoodAbstractExtremeValueModel)

I get the error:

ERROR: MethodError: Cannot `convert` an object of type
  Type{MaximumLikelihoodAbstractExtremeValueModel} to an object of type
  Union{Int64, DataType}

Closest candidates are:
  convert(::Type{T}, ::T) where T
   @ Base Base.jl:84

Stacktrace:
 [1] OutDims(axisdesc::Tuple{}, backend::Symbol, backendargs::@Kwargs{}, update::Bool, artype::Type, chunksize::Symbol, outtype::Type)
   @ YAXArrays.DAT ~/.julia/packages/YAXArrays/ppMtD/src/DAT/registration.jl:102
 [2] OutDims(; backend::Symbol, update::Bool, artype::Type, chunksize::Symbol, outtype::Type, backendargs::@Kwargs{})
   @ YAXArrays.DAT ~/.julia/packages/YAXArrays/ppMtD/src/DAT/registration.jl:134
 [3] top-level scope
   @ REPL[64]:1

It seems to be possible to assign a custom type to outtype, but I am unsure why it fails for MaximumLikelihoodAbstractExtremeValueModel:

struct foo
    a
end

outdims = OutDims(outtype=foo)
OutDims((), :auto, Base.Pairs{Symbol, Union{}, Tuple{}, @NamedTuple{}}(), false, Array, :input, foo)

Thanks for any intput!