The API has an unfortunate combination of purist attitude on some aspects and programmer-friendly type substitution on others.
From a purist geometric point of view, a Point3d is a location in model space. It makes no sense to add two points: what does it mean to add locations? Also in the purist view, a Vector3d is a quantity with direction and magnitude, but no location. When you add a Vector3d to a Point3d the meaning is “create a new Point3d offset by this Vector3d from the original”. You can also add Vector3d’s to each other because geometry defines a triangle-rule meaning for this operation.
But from a programmer-friendly perspective, SketchUp overloads the Array class so that almost any method that accepts a Point3d or a Vector3d will happily accept a 3-element Array as well. This leads programmers to want the Point3d+Point3d operation because the semantic boundary between Point3d and Vector3d and Array has become blurry.