Support for marked point processes

Following the discussion at https://discourse.julialang.org/t/differentialequations-jl-marked-point-processes-or-storing-user-data/19453, I suggest adding support for marked point processes. Specifically, for simulating the process alongside an SDE affected by it, and retrieving times and marks from the solution.

As a suggested interface (slightly different than the one I suggested in the discourse thread above), the constructors for VariableRateJump and ConstantRateJump can have an additional optional (keyword?) argument mark specifying a mark sampling function. The affect! function would then take an additional argument for the mark.

total_rate(u,p,t) = ...
sample_mark(u,p,t) = ...
affect!(du,u,p,t,y) = ...
mpp = VariableRateJump( total_rate, affect!, mark=sample_mark )
prob = JumpProblem( ode_or_sde, mpp )
sol = solve(prob)
times, marks = get_point_events(sol)

which would mean something like

function _affect!(du,u,p,t)
    y = sample_mark(u,p,t)
    store_time_and_mark_somewhere!(t,y,somewhere)
    affect!(du,u,p,t,y)
end
jump = VariableRateJump( total_rate, _affect! )
prob = JumpProblem( ode_or_sde, jump )

Mathematically, this is enough to support equations of the form
du_t=<drift and diffusion terms> +∫_y h(u_t,p,t,y)N(dt,dy)
where N is a MPP with an intensity kernel of the form λₜ(dy)=λ(u_t,p,t,dy) (total_rate is ∫λₜ(dy), sample_mark samples y from λₜ(dy)/∫λₜ(dy))

The issue names VariableRateJump, ConstantRateJump, JumpProblem, solve, and get_point_events but gives no repository file or test path. Start by tracing those constructors and solution event handling; done means marked sampling reaches affect! and times and marks can be retrieved for ODE/SDE jump simulations.