The top is attached (at least in part) using buttons, although it's hard to tell what's going on with the cross-bracing framework. Buttons are one of the established ways of allowing for seasonal changes in the width of a tabletop. These ones are crude though, and as a result may not allow sufficient movement. The ones on the sides in particular must allow for the button to move into and out of their respective slots far enough to account for half of the total movement.
Pictures 3 and 5 seem to show that the buttons could be the cause. In no. 3 the shoulder of the button is hard up against the apron, and in the fifth image it looks like the tongue of the button is a very tight fit in its slot.
The joints are part of the problem
The fact that the tabletop had glue-line failures is another piece of evidence. It indicates that the glued joints are weaker than they should be.
If the glued joints are done properly when there is a movement issue while the tabletop might still split the cracks will completely ignore the joints, because well-made glued joints are literally stronger than the wood around them.
The glue itself or the glueing technique used may be at fault here. Teak is a tropical wood and like many tropical woods it has a certain natural oily/resinous nature which means it can be tricky to glue well. Solvent-cleaning of joint surfaces just prior to glueing is a common recommendation when working with such woods.