ForwardDiff tag error when computing Jacobian for DAE adjoint

[taylormcd]

Here's a MWE:

using OrdinaryDiffEq, LinearAlgebra, SciMLSensitivity, Zygote

function f!(du, u, p, t)
    x, y, vx, vy, T = u
    g, L = p
    du[1] = vx
    du[2] = x*T
    du[3] = vy
    du[4] = y*T - g
    du[5] = 2*(vx^2 + vy^2 + y*(y*T - g) + T*x^2)
    return nothing
end

func = ODEFunction(f!, mass_matrix=collect(Diagonal([1,1,1,1,0])))
u0 = [1.0, 0, 0, 0, 0]; p = [9.8, 1]; tspan = (0, 10.0)
prob = ODEProblem(func, u0, tspan, p)
solve(prob,Rodas5())

function sum_of_solution(p)
    _prob = remake(prob,p=p)
    sum(solve(_prob,Rodas5(),saveat=0.1,sensealg=QuadratureAdjoint(autojacvec=false)))
end

dp = Zygote.gradient(sum_of_solution,p)

ERROR: Invalid Tag object:
  Expected ForwardDiff.Tag{SciMLBase.UJacobianWrapper{ODEFunction{true, typeof(f!), Matrix{Int64}, Nothing, Nothing, Nothing, Nothing, Nothing, Nothing, Nothing, Nothing, Nothing, Nothing, Nothing, Nothing, typeof(SciMLBase.DEFAULT_OBSERVED), Nothing, Nothing}, ForwardDiff.Dual{ForwardDiff.Tag{DiffEqBase.OrdinaryDiffEqTag, Float64}, Float64, 1}, Vector{Float64}}, ForwardDiff.Dual{ForwardDiff.Tag{DiffEqBase.OrdinaryDiffEqTag, Float64}, Float64, 1}},
  Observed ForwardDiff.Tag{SciMLBase.UJacobianWrapper{ODEFunction{true, typeof(f!), Matrix{Int64}, Nothing, Nothing, Nothing, Nothing, Nothing, Nothing, Nothing, Nothing, Nothing, Nothing, Nothing, Nothing, typeof(SciMLBase.DEFAULT_OBSERVED), Nothing, Nothing}, Float64, Vector{Float64}}, ForwardDiff.Dual{ForwardDiff.Tag{DiffEqBase.OrdinaryDiffEqTag, Float64}, Float64, 1}}.