How much of this settling is related to the tire taking a set (flat spotting)?
Whenever I leave the car setting for an extended period or overnight, it is with the car rolled off the scales. That way when the car is rolled back on the scales it is not rolling onto a flat spot caused by sitting in the same spot on the scales. While I frequently roll the car back off the scales to "bounce it after adjustments, I swear by the use of thin greased slip plates under all four wheels and credit their use for the biggest gain in obtaining consistent/repeatable results.
I also set the tire pressures to off-track target pressures to simulate the dynamic tire cross section however I have not been able to detect any measurable cross weight affect from doing this.
I don't think the tire flatspotting matters much, but nothing wrong with eliminating it as a variable. For those that insist on measuring the pinch welds, then tire stagger, "coning", and changes in tire types (Hoosier, Toyo, etc.) will all affect the "ride height" measurement.
The biggest effect I saw on the NB due to "settling" was a big difference in camber, but more worrying - a big difference in front toe due to the hideous bumpsteer. I noticed this overnight, as I would sometimes quit for the night, come back out in the morning and see how much higher my camber #s were.
The NB was always less repeatable on the scales, because MazdaSpeed/AWR left zero compliance in the original '99 SM shock hat setup (I'm not talking about the OEM shock hat setup) - the only "compliance" was the flexing of the body sheet metal and the "bending" (cocking, really) of the shock absorber shaft with respect to its housing - thus turning the shock absorber into a "leaf spring" (except a telescoping leaf spring with a sloppy set of seals in the middle of it). And people wondered why the shocks turned to junk in hours. Or minutes.
In other words, when you took out a lower shock bolt on an NB SM while it was sitting at ride height, the shock absorber would snap outward (or inward, I can't remember) because it was being "bent" due to the lack of compliance/articulation at the shock hat joint. This was SOMEWHAT mitigated by the FatCat hats, depending on how apeshit you go on squishing the rubber isolators, or whether you are running the FatCat supplied material to begin with (ahem).
Note: If you try this shock bolt removal experiment at home, you need to support the car at "ride height" on the corner you take the lower shock bolt out on.