PMP: Spatial sort polygon soup

[afabri]

Summary of Changes

Add CGAL::Polygon_mesh_processing::hilbert_sort_polygon_soup()

Todo

• evaluate gain on higher level functions
• Document in user manual

Release Management

• Affected package(s): PMP
• Small Feature (if any): tbd
• Link to compiled documentation (obligatory for small feature) wrong link name to be changed
• License and copyright ownership: GeometryFactory

[afabri]

I wanted to evaluate the gain, and took tangential_relaxation(mesh) expecting it to be a low hanging fruit. To my surprise it becomes 20% slower, without that I understand yet what happens. As the Hilbert sort sorts the polygon soup
I checked that the vertices./faces/halfedges of the generated Surface_mesh are also sorted. I then was wondering if tangential relaxation depends on the order, but it does not modify positions while iterating over the vertices, but per round.

As a side remark, tangential_relaxation(mesh) uses a std::unordered_map as property map for vertices and faces. This makes sense if it does on a subset of vertices. Otherwise a dynamic property map would be a better choice.

[afabri]

As explained here Hilbert sorting only guarantees that what is close in the sorted sequence is also geometrically close, but not the other way round. I added a benchmrk that only calls faces_around_target() for all vertices.