leonardoa since this is not a bug report, I am moving this to questions category.
Currently you can’t change dimensions of box that surrounds the face. It is calculated automatically based on autodetected face geometry. Faces can intersect, but that should not be a problem as long as both faces were detected correctly. If you have only one face detected, you can detect another one using “Add new face here” function. And it is totally OK if you do that inside other person’s box, that should work anyway. If in the end both faces were detected successfully, there should not be a problem to tag each one. Just point to the center of the box to make sure you select correct face with the mouse.
However there might be cases, when program just can’t correctly find face even if you point it where it is. That happens from time to time for low quality photos, faces that are visible only partially or are looking away from the camera. If that is the case, this face cannot be tagged in the current version.
In the future there will be function that will allow to mark people on the photos even face cannot be detected, or even absent on the photo, but I can’t tell right now anything about dates for this funciton.