mouse look and scan

Is it possible to set up the variables or a set of filters in the sensor function to only report objects in line with the cross hairs in mouse look?

Technically no, for attachments, because they all exist at the center of your avatar and not your face, regardless of where you move them. Even if you set up unattached prims in a way such that a sensor aligned perfectly with your cross hairs, the fact of the matter is that sensors are picky about what boundaries of a target must be within their cone of scanning. For instance, unless an avatar's center (hips) are within the scanned volume, it'll miss them.