Can't use autodiff to calculate jacobian in the time-domain

[heltonmc]

julia> using LightPropagation
julia> using ForwardDiff
julia> ForwardDiff.derivative(dx -> fluence_DA_Nlay_cylinder_CW(1.0, [dx, 0.1], [10.0, 10.0], 1.0, [1.0, 1.0], [1.0, 10.0], 10.0, 0.0, besselroots[1:1000]), 0.1)
-0.12899771300257823

julia> ForwardDiff.derivative(dx -> fluence_DA_Nlay_cylinder_TD(1.0,1.0, [dx, 0.1], [10.0, 10.0], 1.0, [1.0, 1.0], [1.0, 10.0], 10.0, 0.0, besselroots[1:1000]), 0.1)
ERROR: MethodError: no method matching ldexp(::ForwardDiff.Dual{ForwardDiff.Tag{var"#11#12", Float64}, Float64, 1}, ::Int64)
Closest candidates are:
  ldexp(::Float16, ::Integer) at math.jl:787
  ldexp(::T, ::Integer) where T<:Union{Float16, Float32, Float64} at math.jl:745
  ldexp(::BigFloat, ::Int64) at mpfr.jl:626
  ...
Stacktrace:
  [1] sqrt(z::Complex{ForwardDiff.Dual{ForwardDiff.Tag{var"#11#12", Float64}, Float64, 1}})
    @ Base ./complex.jl:488
  [2] α_coeff!
    @ ~/.julia/packages/LightPropagation/nHQWE/src/forwardmodels/Diffusion Approximation/DAcylinder_layered.jl:309 [inlined]

This is caused by the inverse laplace transform needing to use complex variables. The error is thrown because we need to evalue sqrt(Complex{ForwardDiff.Dual{Float64}... unsure the best fix for this. We are ok when calculating the flux because the z variable is never square rooted

For references the function throwing error is:

@inline function α_coeff!(α, μa, D, sn)
    @inbounds for ind in 1:length(μa)
        α[ind] = sqrt(μa[ind] / D[ind] + sn^2)
    end
    return α
end

[heltonmc]

This seems to work now on v0.9.0 but is incredibly slow.... perhaps linked to #21

julia> ForwardDiff.derivative(dx -> fluence_DA_Nlay_cylinder_TD(1.0,1.0, (0.1, dx), (10.0, 10.0)), 0.1)
-0.002803585371894059