new triangulation algorithm based on sweepline and retriangulation to tackle the current algorithm problems

[jimy-byerley]

This issue is to track advance in experimenting this particular algorithm. Its targets:

• being faster than the current algorithm (meaning should not have a $O(n^2)$ complexity), the current algorithm is alone responsible of the computation time of boolean operations
• being robust to all kind of degenerated and side cases
• output nice triangulations (with triangles the closest to equilateral triangles) so avoid loosing computation precision in operations on the resulting mesh

This should solve #79, #54 once it is done

The principle:

• [x] triangulate using sweepline algorithm $O(n log(n) )$
• [ ] rework the triangulation by propagating better pairing candidates $O(n)$

[jimy-byerley]

the sweepline triangulation seems to work

[jimy-byerley]

possible degenerated cases for sweepline

• [x] fix degenerated case of 2 edges at the same place
• [ ] fix degenerated case of a null-length edge
• [x] fix degenerated case of 2 monotonics with the exact same flat front, with one getting challenged by an edge of the other
• [x] fix degenerated case of empty loop
• [ ] fix degenerated case of split point on an edge