Have read the referenced study and have a few comments.
The study references both diffraction and sound impedance of the enclosure for the speaker, but then creates enclosures where the volume of the enclosure and the distance of the speaker/piston to the furthest rear surface varies quite a bit, from 2 foot to 6 inches. Without going through the full math of the impedance and reflection pressure on the speaker/piston for each shape used, the study is only partially conclusive.
Since the study does not include reference to any sound dampening material used within any of the enclosures, we have to assume that there is none. This should be required information for complete and accurate understanding of the results.
The material used for the various shapes, and the construction methods used are not detailed, so the shape information is only partially informative. The absorption and reflection properties of the material used and joining methods will affect the diffraction and impedance of each enclosure.
There is no charting/graph of the sample speaker free standing (no enclosure) for reference, so the charts graphs can only the used as comparison to each other. We do not know what the free air response of the sample speaker is for comparison to the enclosures.
The shapes/volume of enclosures for the size of speaker used would mostly argue for freestanding speakers. A speaker of 7/8 inch diameter (a tweeter) requiring a sphere of 2 foot diameter for a relatively flat response or a cubic volume of over 2 feet for a similar response would appear to require a massively sized enclosure for a full range speaker that would include at the minimum a second driver of at least 5 1/4 inch diameter. Thus one would conclude that freestanding speakers with a bare minimum of mounting surface would be best.
What the study does show is that enclosure volume will affect response curve. Distance from speaker to rear surface will affect response curve. Internal shape of enclosure will affect response curve. And the study hints that minimal front mounting surface may be beneficial to the response curve, but is confused by the various internal shapes and volumes to provide a clear conclusion.
Most all of this is known, and most all high end speakers use cabinet shaping, material, volume, and dampening for each individual driver to optimize the response curve of each driver, and the response curve of the speaker as a whole as a single point sound source. For those whose room volume, configuration and dampening are optimal then full planar or freestanding speakers can be the best, otherwise it is all compromise.