Not entirely true, the spall liner doesn’t HAVE to be the last layer. To my knowledge, it DOES have to be bonded to the last layer(s). A good example is laminated glass’ it is typically arranged in a basic glass-interlayer-glass bonded together arrangement. Upon a bullet passing thru, the last layer remains attached to the interlayer and there is no spalling besides where the bullet penetrated in a narrow cone.
As for the Kevlar inserts being what is typically mentioned, I do have my doubts about it being Kevlar. Kevlar can be a bit tricky to bond to with metal and it also would depend on how thick the metal is too. For example, if the metal is too thick and cannot plastically deform enough with the kevlar, then it will reach it’s ultimate strength and create spalling instead of simply deforming. But if it is thinner, it will be able to plastically deform and the bond will hold and prevent spalling. A bonding agent like this: Supreme 11HT Product Information | MasterBond.com could be used, but it would be a bit tedious depending on how they manufacture it. If I still had my ANSYS commercial license I would’ve been tempted to run this simulation to see how spalling changes without a liner, with a liner, and with a bonded liner. But if Gaijin truly wants an answer they can afford a ANSYS license and can hire an FEA engineer to run it.