Low Flier, what makes you think they are looking for the box? They first have to develop a notion of where to look with high resolution or even optics. They look for larger pieces of the aircraft, such as the engines, first. Then they start looking for the little stuff within an established debris field.
In round numbers the speed of sound in water is 1500 m/s. 120kHz gives a resolution of about 1500/120000/2 (half wavelength) or about 6 millimeters. Anything smaller than that will be hard to resolve. Something 10 times larger should be within the realm of detection without really fancy processing if it's outside the range of sizes of uninteresting objects in the area.
Off hand I'm curious where you got the sonar resolution data you cited. I presume sonar is different from radar, where I do have some engineering know-how. I also do note that the radar problem is seldom trying to detect an object sitting on another object or the ground. I'm willing to learn.
{^_^}