It seems like Feature Geometry doesn't make reference to Archiphonemes - only Underspecified Segments. Is there a theoretically-motivated reason for this, if it is true?
Trubetzkoy's archiphoneme theory had no reference to underspecified features, since it was proposed in a day before feature theory existed. The interpretation of archiphonemes as being underspecified segments is ahistorical. Most of Trubetzkoy's case studies of archiphonemes can reasonably be interpreted as involving underspecified segments, but there is no theoretical guarantee that they all can.
Maybe FG people want to make clear that their theory is more constrained than Trubetzkoy's (which I think would be correct). Or maybe they are just being fastidious and do not want to attribute to Trubetzkoy something he didn't actually propose.