I'm quite new with cgal but for experience: Yes, but it will answer false to cdt.is_valid

You'll end up with a constrained delaunay triangulation where all the constraints are on T1, but a triangulation nonetheless (which doesnt satisfy the delaunay property).

If some of the points fell inside the T1 region you might get new triangles there, with the T1 edges untouched, but apparently that's not the case for you.

Note that T2 will connect with closest vertices of T1.