Add support for random time span within trajectories

[utkarsh530]

Fixes #141 MWE:

using DiffEqGPU, OrdinaryDiffEq, StaticArrays, LinearAlgebra, CUDA
function lorenz(u, p, t)
    σ = p[1]
    ρ = p[2]
    β = p[3]
    du1 = σ * (u[2] - u[1])
    du2 = u[1] * (ρ - u[3]) - u[2]
    du3 = u[1] * u[2] - β * u[3]
    return SVector{3}(du1, du2, du3)
end

u0 = @SVector [1.0f0; 0.0f0; 0.0f0]
tspan = (0.0f0, 10.0f0)
p = @SVector [10.0f0, 28.0f0, 8 / 3.0f0]
prob = ODEProblem{false}(lorenz, u0, tspan, p)

saveats = 1.f0:1.f0:10.f0

monteprob = EnsembleProblem(prob, prob_func =  (prob, i, repeat) -> remake(prob; tspan = (0.f0, saveats[i]), saveat = LinRange(0.f0,saveats[i],10), safetycopy = false))

alg = GPUTsit5()

sol = solve(monteprob, alg, EnsembleGPUKernel(0.0), trajectories = 10,
            adaptive = false, dt = 0.01f0, save_everystep = false)

[utkarsh530]

@ChrisRackauckas does this seem okay? We'll need to document it a bit because I know people will break this easily haha

[ChrisRackauckas]

Documentation is only half the battle. Make it throw clear and descriptive errors before it hits the kernel for the cases that will not work. There's no error exceptions thrown with this at all right now.

[utkarsh530]

Tried to throw exceptions for common cases.