vcat/hcat/cat bug at edge case with one KeyedArray

[mzgubic]

julia> ka = wrapdims(rand(3), obs=1:3)
1-dimensional KeyedArray(NamedDimsArray(...)) with keys:
↓   obs ∈ 3-element UnitRange{Int64}
And data, 3-element Vector{Float64}:
 (1)  0.16740473902785014
 (2)  0.17521465963627714
 (3)  0.05600551386851893

julia> vcat(ka, ka) # all good
1-dimensional KeyedArray(NamedDimsArray(...)) with keys:
↓   obs ∈ 6-element Vector{Int64}
And data, 6-element Vector{Float64}:
 (1)  0.16740473902785014
 (2)  0.17521465963627714
 (3)  0.05600551386851893
 (1)  0.16740473902785014
 (2)  0.17521465963627714
 (3)  0.05600551386851893

julia> vcat(ka) # turns into a NamedDimsArray
3-element NamedDimsArray(::Vector{Float64}, :obs):
↓ obs  0.16740473902785014
       0.17521465963627714
       0.05600551386851893

julia> hcat(ka) # turns into a Matrix
3×1 Matrix{Float64}:
 0.16740473902785014
 0.17521465963627714
 0.05600551386851893

julia> cat(ka; dims=1) # turns into a Vector
3-element Vector{Float64}:
 0.16740473902785014
 0.17521465963627714
 0.05600551386851893

[mcabbott]

That's an oversight, these loops

AxisKeys.jl/src/functions.jl

Lines 121 to 128 in f172d7a

	for (T, S) in [(:KeyedVecOrMat, :KeyedVecOrMat), # KeyedArray gives ambiguities
	(:KeyedVecOrMat, :AbstractVecOrMat), (:AbstractVecOrMat, :KeyedVecOrMat),
	(:NdaKaVoM, :NdaKaVoM),
	(:NdaKaVoM, :KeyedVecOrMat), (:KeyedVecOrMat, :NdaKaVoM),
	(:NdaKaVoM, :AbstractVecOrMat), (:AbstractVecOrMat, :NdaKaVoM),
	]
	@eval function Base.vcat(A::$T, B::$S, Cs::AbstractVecOrMat...)

and the similar ones for cat don't include the one-argument case.

[mzgubic]

I can make a PR this week if you'd like?