It looks to me like the multi-element wing is key here.
When flow is attached, we get flow on this general order:
To me, it looks like it's key that most of the air flowing over the bottom side of the leading element also flows pass the bottom side of the second and third elements, so it's increasing down-force the whole way. We get down-force (and induced drag) somewhat like we would from a single, heavily cambered, airfoil--other than the minor detail that a single airfoil like that would most likely stall.
But then let's consider what happens in a stall. We now have flow separation and cavitation:
Because of the flow separation, the air that flowed past the leading element is no longer directed upwards, so it no longer flows past the second and third elements (and likewise, air that flowed over the second element no longer flows past the third.
So, instead of acting like a single airfoil, it now acts like three completely separate airfoils and (key element) most of the air only flows past a single element instead of all three.
Apparently the increased induced drag from air flowing over all the elements is (probably only slightly) more than the increase in drag from being stalled.
Image Source
As an aside: although it doesn't talk specifically about what happens when the wings stall, you might find this paper interesting in any case (if you haven't already read it, of course).
https://hrcak.srce.hr/file/238785