O-ing, I loaded
your gps track from your AP log onto Routegadget, using the "splits-based" calibration option, mainly so I had a cross-check on the event load. The original gps looks like it misplaced by up to 50m, but the RG routine seems to pull it into something semi-believable, allowing for the big building effects.
Any comments on how well RG has adjusted the track?