The tensile strength is often related to the "pull out strength" of the rivet. Rivets will almost always fail in shear first but there are cases where the rivet being used is tougher than the plate and the plate will crush or tear(thick steel rivet and thin aluminum plate for example). It usually requires more strength to pull a rivet out than to shear it which is what the datasheet is letting you know. The reason you do not use a rivet for a tensile example is because you rarely use one rivet in an application so adding multiple rivets makes the force to shear in that joint increase whereas the pullout force for that ONE rivet still remains about the same no matter how many you add because you are only relying on the head of the rivet and the friction in the expanded joint to hold it in. If separation occurs then shear will most likely happen and cause failure.
Joe is spot on about dynamic and cyclical loads where vibration or fully reversed loading occurs. These types of loading conditions in the real world are treated with a high factor of safety (6 to 12 times) over static loads and more often than not are exactly what you experience on a robot during competition.
Our robot chassis this year was all riveted construction with only two joint welds.
http://www.yetirobotics.org/wp/our-robot/
We sheared the riveted angle plate that stops our gravity claw twice during competition so we added more rivets to fix it at worlds since it was a 2 minute fix. Once one rivet shears it usually unzips and breaks the rest of them soon after as the kids got to see.
Hope that helps!