While not remotely scientific, my testing (using an TMA40-4 Anemometer) (with and without the HEPA) showed a reduction of output CFM with the baffle as compared without. My summation of the results at the time was more air was being forced through the carbon pellets.
While I've not tested your revised design, I suspect an even greater amount of bypass. Your design doesn't appear to incorporates, (to help facilitate ease of printing and part reduction) any chambered area between the top grate to guarantee carbon passage. My original design with the extension incorporates a 20.22mm carbon chamber where the air must pass through the carbon before exiting. My thinking at the time was primarily to help control VOC's
Perhaps the pellets I'm using are of a smaller diameter and length and thus cause greater bypass.