Make ODEForwardSensitivityProblem work with scalar `u0`

[ChrisRackauckas]

MWE:

using SciMLSensitivity, DifferentialEquations

tspan = (0.0,1.0)
function parameterized_lorenz!(du,u,p,t)
    x,y,z = u
    σ,ρ,β = p
    du[1] = dx = σ*(y-x)
    du[2] = dy = x*(ρ-z) - y
    du[3] = dz = x*y - β*z
  end

function simulate_toymodel(p)
    u0 = [1.0,0.0,0.0]
    prob = ODEProblem(parameterized_lorenz!,u0,tspan,p)
    sol = solve(prob)
end

p = [10.0,28.0,8/3]
Twindow = 0.2

input(t, p) = simulate_toymodel(p) |> sol -> sol(t; idxs=2)
f!(u, p, t) = (input(t, p) -  u) / Twindow
u0 = mean(input(0:Twindow, p))
prob_simple = ODEProblem(f!, u0, tspan, p)
sol_simple = solve(prob_simple, Vern9(); abstol=1e-6, reltol=1e-6 ) # OK

prob = ODEForwardSensitivityProblem(f!, u0, tspan, p) # Error
sol = solve(prob, Vern9(); abstol=1e-6, reltol=1e-6)