Airframes are generally pretty low margin
Composites do indeed fail catastrophically when they go, but that's not to suggest that they can't be designed with sufficient margin. While there are some intricacies to composite design, the basic concepts of stress-versus-load still apply.
While you're correct that metal has more capability for plastic (yielding) deformation before failure, it's also the case that aircraft are manufactured right down to the razor's edge of margin. Every extra pound means you have to burn that much more fuel on every trip. I think that the 747's cross-section area between the frames is only something like twelve square inches. So while the metal could conceivably have some strength left after yielding, it probably wouldn't have all that much; and aluminum doesn't yield all that much before failure anyway (only a factor of 1.12 between yield stress and failure stress, versus about 1.67 for cold-drawn stainless steel.)
If you're worried about composites being used as major structures, that ship has sailed (er, that plane has flown?) Airbus has been making composite tails for years--in fact, in November 2001 there was an airliner crash in New York where the tail snapped off due to overstressing (pilots had been improperly trained to seesaw the rudder violently to handle turbulence.)
Ultimately, it seems like it's as you describe--the benefit of composites (or weldments) is that you can reduce the amount of joinery you need, saving that part of the weight. You also get a small benefit from reducing the work at your primary assembly facility (bigger subsections = less time bolting them together) although you're really just spreading the labor out to the subsection manufacturers.
Note that they're still using fasteners to assemble the 787 (in fact there was a big problem with their fastener supply chain--they just assumed that there would always be plenty of aircraft-grade bolts available. Unfortunately, when all the defense-industry work went away, the fastener suppliers went out of business!) They just have to be more careful about it. i.e. instead of just punching a hole with a simple die punch, they have to drill and ream and clean. Instead of just slapping on a rivet and letting everything mush itself into place by yielding, they have to use a special-made washer with a curved face to avoid gouging the composite surface (or spend time countersinking every place there's going to be a fastener.) People have tried to make bonded joints for composite structures and nobody's ever found a way to do it that's as good as a bolted joint without being just as much work. (you have to design specifically to make the bonded joint work--any peel stress will kill it dead, just like ripping a piece of Velcro open versus trying to slide it off.)