I just received the file and ran it in two versions of GPS-Results. They both get the same results as you got from GPS-SS. Of course, this is to be expected because GPS-SS use the program in the website.

But IMHO, the results from the 2 seconds is highly suspect and unreliable. There are huge sawtooth spikes in the speed graph from second to second on many parts of the track.
The best 10 second actually looks quite OK on the graph so it is probably quite representative. Some of the other 10 sec results have the big spikes in them so you 5 x 10 sec is probably less indicative.
Screenshot of sawtooth graph:
I'll email you some more observations after I have gone for a sail.