In the case of dolphin sonar, there is one respect in which it is pretty straightforward: the advantage of each minute improvement in the dolphin’s hearing/sonar abilities would be immediately relevant to its welfare. Even human hearing is sufficient for a certain basic level of echolocation (see the blind skateboarder) and the basic traits of being able to make high-pitched noises and honing a sense of hearing for hunting in a watery environment need no special pleading.
Keep in mind too that the circuitous path taken to the dolphin’s present-day excellence may not represent the best route it could have taken, just the route it did take. You don’t want to fall into the trap of seeing something complicated and believing intricacy is a sign of intelligence. Sometimes true genius is in simplicity.
But I do know what you mean, I’ve stared at my biology textbooks and wondered how…? I think one big difference between new technology and new biology is that technology tends to be very feature-focused—how can I make it bigger, faster, add more bells and whistles? While those things almost wind up being incidental to the massive part of the work that goes on biologically, most of which is about making us robust to whatever unexpected things happen to a creature. Damage repair, redundancy so that losing the use of one eye or ear or leg doesn’t crash the system, making sure we can survive famine or falls or swim to shore. And most of that work was done long before we started tottering around on two legs.
That’s why I never think it’s weird that evolution took so long perfecting a single cell before multicellular life expanded, or spent so long with just insects or fish or lizards or anything else. It only makes sense that all the stuff under our hood that we take for granted would take the better part of the history of the world to get right.