The short answer is "no idea" and it's not guaranteed that what you see if the same as what I see. On my LE13 image there's 100% a bad leak, but I'm running a whole heap of experimental things that mess with DRM planes and buffers and I see the same leak on Amlogic and Rockchip (with the same patches) too. The fact nobody but you has complained before on LE12 is the main thing I find odd, as it's the kind of thing that would attract forum complaints.
Either way, the LE13 patches need a load of reviewing and debugging and that's going to be my (Claude driven development) focus over the next week. Maybe that results in something..