The CAP a-arms have the rod ends threaded into the tubes. In order for the tube to be strong enough to support threading in the rod end it has to be MUCH thicker than it would need to be to just support the normal A-arm loads. This being the case, the tube section of that A-arm is overbuilt. The weld, on the other hand, is basically the same size as it would be if the tube were of a thinner wall. The strength of the weld depends on its thickness and length. The length can't change in this application unless you go with a much larger diameter tube, which isn't practical. The thickness isn't just the depth, its the width of the weld, which is also fairly limited for this application. The depth can be larger with the increased thickness of the tube, but its a fairly small increase in the strength of the weld compared to the increase in the strength of the tube itself.
So, you have a short, thick wall section of tubing welded to a short, thick section of ball-joint retainer. That will NOT act like most members involved in a weld. For a long section of thin wall tube, yes, the tube should have bent before the weld broke. But for this instance that may not be the case at all, the bending loads required to bend a short section of thick wall tube could very well be higher than the shear load needed to break the weld. Which is exactly why the joint should be gusseted, to add inches of weld and thereby strength to the joint.
If a weld joins two 6 inch sections of solid tubular rod should those bend first too? There are a lot of variables to consider, obviously at one extreme of the example the tubes should bend, at the other extreme, the weld will break first. To say that the tube should bend first in every single case, and every single possible loading condition just isn't accurate.
