Declare that an `AbstractDimStack` is an `AbstractArray` with namedtuple eltype?

[tiemvanderdeure]

After #811 merged, any DimStack now behaves a lot like an AbstractArray that iterates NamedTuples. Should we declare it as such?

It could even be an AbstractBasicDimArray, but that would require some changes as a AbstractDimStack does not have its dimension in its type, currently.

It would allow us to remove a whole list of array-like methods, and make additional things work, e.g.:

A = [1.0 2.0 3.0; 4.0 5.0 6.0]
dimz = (X([:a, :b]), Y(10.0:10.0:30.0))
da1 = DimArray(A, dimz; name=:one)
da2 = DimArray(Float32.(2A), dimz; name=:two)
da3 = DimArray(Int.(3A), dimz; name=:three)

s = DimStack((da1, da2, da3))

broadcast!(s, s) do x
    (one = 1, two = 2, three = 3)
end

This now fails because ndims is not implemented for DimStack (an oversight, I assume), but if I just add
abstract type AbstractDimStack{K,T,N,L} <: AbstractArray{T,N} end
then it just works.

[rafaqz]

Yeah we are getting closer. There's an old PR already for this

(There may still be some problems where we break the interface in some places)

[tiemvanderdeure]

I can't find that old PR but I think we're getting really close. It will be a bit of work cleaning up all the method ambiguities, though.

Maybe we could try to see how hard it is to get tests to pass?

[rafaqz]

Yeah weird I cant either. I'm sure I tried that.

But yes just adding the supertype and running tests would be interesting.