Pure speculation on my part but I assumed Martin came up with the belly bridge design first and used the bottom belly approach. Gibson, recognizing the superior adhesion properties from the larger surface area, but not wanting to copy Martin (or maybe to avoid patent infringement), just flipped it around.
From a structural standpoint, the bottom belly approach makes more sense to me since there is more area glued down behind the saddle to react against the strings as they try to lift up the bridge. I have no data to suggest there are any structural issues with either approach assuming a properly glued bridge. Soundwise, I seriously doubt there is any difference.