When I opened Model_1, I get the following results from restraint summary for case 2.

@432
-20857 -70238 -2997 -1705 0 11865

@433
-23730 -70238 -3409 0 0 0

The resultant FX and FZ are as follows:
@432 = 21071
@433 = 23974

So the Resultant at 432 seems to match FY * 0.3 friction coefficient. But the element forces for 434-432 match the magnitudes of the loads at 433.

I started playing with the Friction Normal Force Variation in the config editor. I lowered the value until 0.001 and I noticed that the FZ and FZ of both nodes start to converge to the common values. Shown below are the Friction Normal Force Variation setting versus the percent difference between the FX of nodes 432 and 433 (also applicable to FZ).

0.15 - 13.77%
0.05 - 0.59%
0.001 - 0.01%

So with 0.001, I get the following values of FX, FY and FZ
@432
-20843.266 -70256 -3129.442

@433
-20840.896 -70256 -3129.087


I'm not sure of the exact mathematical explanation for it. Help file says the Frictional Normal Force Variation is "the amount of variation in the normal force that is permitted before an adjustment is made in the sliding friction force." So we're tightening the tolerance, forcing the solver to iterate enough to meet the tolerance?