Delauney triangulation: Add concave hull as constraint

[sito115]

The function delaunay_2d_cdt currently triangulates the mesh based on the convex hull of the input point cloud. For example, the following point cloud is triangulated as shown below:

To better capture non-convex boundaries, I introduced an additional concave hull constraint, which produces the following result:

This approach is based on the Shapely implementation of concave hull computation and therefore introduces an optional additional dependency. The availability of this dependency is checked at runtime.

The operator tesselation.delaunay now accepts two additional input arguments, following the
Shapely concave hull API

• alpha_concave_hull
  A number in the range [0, 1]. Higher values include fewer vertices in the hull and approach the convex hull. If a negative value is provided, the default convex hull is used.

• allow_holes (bool)
  If set to True, the resulting concave hull may contain interior holes.

The approach has been tested with my dataset; however, additional testing across a broader range of inputs is still required.