If feasible, computes the optimal cost of the entire model with regards to
the optimizer_core_'s dimension costs, minimizing cumul soft lower/upper
bound costs and vehicle/global span costs, and stores it in "optimal_cost"
(if not null).
Returns true iff all the constraints can be respected.
If feasible, computes the optimal values for cumul and break variables,
minimizing cumul soft lower/upper bound costs and vehicle/global span
costs, stores them in "optimal_cumuls" (if not null) and optimal breaks,
and returns true.
Returns false if the routes are not feasible.
Similar to ComputeCumuls, but also tries to pack the cumul values on all
routes, such that the cost remains the same, the cumuls of route ends are
minimized, and then the cumuls of the starts of the routes are maximized.