The problem with the MS profile, of which the example you quote is typical, relates to
(a) asymmetric loading, causing failure of the button or the track lip. An Australian design engineer, Rudi Paspa, many years ago patented an interesting and quite effective way of overcoming this problem by a redesign of the attach button but it never took off for whatever reason. The present dynamic standards, requiring a test simulating floor disruption, have generated a standard method of overcoming this problem to some extent, by incorporating a flexible link in the lower leg to permit the button some degree of rotation so that the button flanges continue to load both track lip faces.
(b) minor deflection of the seat structure, permitting the button to move into the cutout space and leave the track. This can be overcome by having individual shear pegs (to prevent movement by locking the leg into the track) or by having a reasonably rigid seat keel to lock a button to a remote shear peg.