The result of this triangulation will be a planar triangulation, i.e., each point will be a vertex of some triangles (and only triangles). So, as you pointed out, this just connects a set of points without considering any topological constraint (i.e. without assuming they represent the boundary of a polygon. Such a triangulation would be called a constrained triangulation because we would see the polygon edges as constraints between points).
The implemented algorithm is incremental, i.e., each point provided is added one-by-one to an existing triangulation and it is ensured that this point does not lie on any other existing triangle's circumcircle. But this circumcircle property is valid only when the point is added, so it might be violated later when other points are added. In other words, the result will be a topologically well-formed triangulation (and with triangles that should not be too disproportionate), but it won't be a Delaunay triangulation unfortunately (which I would like to add in the future).