spectral_mixture_kernel constructor is not type-stable

[theogf]

Here is a MWE:

julia> v = ones(2);
julia> rbf = rand(2);
julia> cos = rand(2);
julia> spectral_mixture_kernel(v, rbf', cos')
Sum of 2 kernels:
        Product of 2 kernels:
                        Squared Exponential Kernel (metric = Distances.Euclidean(0.0))
                                        - Linear transform (size(A) = (1, 1))
                                        - σ² = 1.0
                        Cosine Kernel (metric = Distances.Euclidean(0.0))
                                        - Linear transform (size(A) = (1, 1))
        Product of 2 kernels:
                        Squared Exponential Kernel (metric = Distances.Euclidean(0.0))
                                        - Linear transform (size(A) = (1, 1))
                                        - σ² = 1.0
                        Cosine Kernel (metric = Distances.Euclidean(0.0))
                                        - Linear transform (size(A) = (1, 1))

julia> @inferred spectral_mixture_kernel(v, rbf', cos')
ERROR: return type KernelSum{Tuple{KernelProduct{Tuple{ScaledKernel{TransformedKernel{SqExponentialKernel{Distances.Euclidean}, LinearTransform{LinearAlgebra.Adjoint{Float64, SubArray{Float64, 1, LinearAlgebra.Adjoint{Float64, Vector{Float64}}, Tuple{Base.Slice{Base.OneTo{Int64}}, Int64}, true}}}}, Float64}, TransformedKernel{CosineKernel{Distances.Euclidean}, LinearTransform{LinearAlgebra.Adjoint{Float64, SubArray{Float64, 1, LinearAlgebra.Adjoint{Float64, Vector{Float64}}, Tuple{Base.Slice{Base.OneTo{Int64}}, Int64}, true}}}}}}, KernelProduct{Tuple{ScaledKernel{TransformedKernel{SqExponentialKernel{Distances.Euclidean}, LinearTransform{LinearAlgebra.Adjoint{Float64, SubArray{Float64, 1, LinearAlgebra.Adjoint{Float64, Vector{Float64}}, Tuple{Base.Slice{Base.OneTo{Int64}}, Int64}, true}}}}, Float64}, TransformedKernel{CosineKernel{Distances.Euclidean}, LinearTransform{LinearAlgebra.Adjoint{Float64, SubArray{Float64, 1, LinearAlgebra.Adjoint{Float64, Vector{Float64}}, Tuple{Base.Slice{Base.OneTo{Int64}}, Int64}, true}}}}}}}} does not match inferred return type Any
Stacktrace:
 [1] error(s::String)
   @ Base ./error.jl:35
 [2] top-level scope
   @ REPL[56]:1

[theogf]

Now that I think about it, it makes sense. There is no way the size of the mixture can be inferred unless we would pass some static vectors or tuples.