conv fails for complex inputs

[ericphanson]

julia> x = rand(5) + im*rand(5);
julia> y = Constant(rand(5));
julia> conv(x,y)
ERROR: InexactError: Float64(0.43372749030578417 + 0.16456075616898147im)
Stacktrace:
 [1] setindex!(::Array{Float64,1}, ::Complex{Float64}, ::Int64) at ./complex.jl:37
 [2] SparseMatrixCSC{Float64,Int64}(::Array{Complex{Float64},2}) at /Users/osx/buildbot/slave/package_osx64/build/usr/share/julia/stdlib/v1.1/SparseArrays/src/sparsematrix.jl:394
 [3] convert at /Users/osx/buildbot/slave/package_osx64/build/usr/share/julia/stdlib/v1.1/SparseArrays/src/sparsematrix.jl:430 [inlined]
 [4] _to_same_csc at /Users/osx/buildbot/slave/package_osx64/build/usr/share/julia/stdlib/v1.1/SparseArrays/src/sparsematrix.jl:2566 [inlined]
 [5] setindex!(::SparseMatrixCSC{Float64,Int64}, ::Array{Complex{Float64},1}, ::UnitRange{Int64}, ::Int64) at /Users/osx/buildbot/slave/package_osx64/build/usr/share/julia/stdlib/v1.1/SparseArrays/src/sparsematrix.jl:2573
 [6] conv(::Array{Complex{Float64},1}, ::Constant{Array{Float64,1}}) at /Users/ephhome/.julia/dev/Convex/src/atoms/affine/conv.jl:16
 [7] top-level scope at none:0

The problem is the line https://github.com/JuliaOpt/Convex.jl/blob/fe366821b77d8243f2ca21a7deb35f4d50c579ce/src/atoms/affine/conv.jl#L14 where a Float64 matrix is constructed to hold elements of x.