If the mirror is monolithic, why wouldn't back plate deformations be transmitted to the front through the spacer posts in between?
Because it's not monolithic. I have made a crude drawing below. The big black circle are the two mirror plates (we are looking straight down). The blue squares are the posts between the plates (and there are probably more in the real thing, but I got tired of drawing them...). The 6 red circles are the 6 cell supports.
So let's say that the back plate is deformed upward by say a full wave by the 6 cell supports. The deformation at the nearest post will not be nearly as large, and would mostly be a pull towards the cell point on the order of a fraction of a wave. The deformation transferred to the front plate would be a product of the angle of the tiny sideways movement of the base of the plate over the length of the post (fraction of a wave divided by inches).
Michael, of course you are doing what the customer asks. I think this would be more a job for the mirror maker to test. I suspect that cellular designs can handle much simpler cells than monolithic mirrors can. It should be easy enough for the mirror maker to do a test (I presume he has testing equipment for making the mirror anyway). Put the mirror on a 27 point cell, take an interferogram, then put it on a 6-point cell and repeat.
From the mirror maker standpoint, simpler cell requirements could be a selling point, and partly offset the additional cost of the cellular blank with savings on the cell.