Python: Add support for sweep arranger for the ROUTING_SWEEP strategy.

[NathanNguyen-Dev]

I am trying to set First Solution as SWEEP to be able to get the optimizer running with cluster of nearby drop point instead of having expanding tree.

 # Setting first solution heuristic.
    search_parameters = pywrapcp.DefaultRoutingSearchParameters()
    search_parameters.first_solution_strategy = (
    routing_enums_pb2.FirstSolutionStrategy.SWEEP)

    # # Setting first solution heuristic.
    #     routing_enums_pb2.FirstSolutionStrategy.PATH_CHEAPEST_ARC)
    search_parameters.local_search_metaheuristic = (
        routing_enums_pb2.LocalSearchMetaheuristic.GUIDED_LOCAL_SEARCH)

However, I encounter this error
Invalid RoutingSearchParameters for this model: Undefined sweep arranger for ROUTING_SWEEP strategy.

Is there another way to have cluster of point server by the same route ?

[Mizux]

Seems to be C++ only and need to define and use the SweepArranger object.

or-tools/ortools/constraint_solver/routing.h

Lines 1398 to 1403 in c08b48d

	#ifndef SWIG
	// TODO(user): Revisit if coordinates are added to the RoutingModel class.
	void SetSweepArranger(SweepArranger* sweep_arranger);
	/// Returns the sweep arranger to be used by routing heuristics.
	SweepArranger* sweep_arranger() const;
	#endif

or-tools/ortools/constraint_solver/routing_search.h

Lines 1226 to 1242 in 92eab40

	/// Class to arrange indices by their distance and their angle from the depot.
	/// Used in the Sweep first solution heuristic.
	class SweepArranger {
	public:
	explicit SweepArranger(
	const std::vector<std::pair<int64_t, int64_t>>& points);
	virtual ~SweepArranger() {}
	void ArrangeIndices(std::vector<int64_t>* indices);
	void SetSectors(int sectors) { sectors_ = sectors; }
	private:
	std::vector<int> coordinates_;
	int sectors_;
	DISALLOW_COPY_AND_ASSIGN(SweepArranger);
	};
	#endif // SWIG