I can understand your frustration, but it's time to stop ranting and just deal with the issue. Posting here does not deal with the issue very effectively, unless your intention is just to bash Haas. No company is perfect and no piece of equipment is perfect. Overall, those of us who have our Haas machines are very happy.
I do not know what the spec's are on the built in rotary in your machine, but I looked up the spec's for an HRT210 and it was 30 arc-sec of backlash. That does not seem like much, but it is 120th of an arch minute or .0083 degrees.
Over the 28 inches you say you are trying machine, that equals .0041" of deviation. Then you can add on the accuracy and repeatability of the unit. It makes sense to me to do as recommended and remove the backlash with the procedure recommended.
On the ways of the machines the recirculating ball nuts can take the pressure because they are always rolling and not rubbing, but on a rotary axis the is always rubbing between the worm and the gear. If you make the pressure too great, in order to remove the backlash, the gears will fail. Then of course you have other connections between the encoder and the actual table surface.
I hope that you get the issue resolved, but I think that setting it up by dialing it in and setting the A axis and not some parameter would be more accurate and repeatable.
Best of luck with your issue.
Mike
Two Haas VF-2's, Haas HA5C, Haas HRT-9, Hardinge CHNC 1, Bother HS-300 Wire EDM, BobCAD V23, BobCAD V28