Some additional information

1. There are alternative rules in Case 2286 (it seems that Rev6 is the last one, so it's 2286-6), which are intended to deal with the excessive conservatism in the original ASME methodology. In PV-Elite the Case 2286 is supported but I haven't accurate information which Case revision because I have access to PV-Elite few revisions old.

2. An old book, Brownell and Young- Process Equipment Design Handbook has a chapter that explains the roots of the method. For me, it seems that today the safety factor between theoretical pressure and allowable pressure is a factor of 3, not 4 as in the original work, but this is just a detail.
One can understand that A is in fact f/E where f is stress, A is strain, E is the modulus of elasticity.
The CS-2 curve is just an attempt to calculate B=f/2 when you know A=f/E. Maybe for us, today, the graph CS-2 is rather confusing but in fact is just the stress-strain graph (scaled down with a factor of 1/2) considering also the influence of temperature on E. Anyway, the shape looks like steel stress-strain non-linear idealized curves, no doubt. What today seems to be strange is the fact graphs are not drawn for small strain, that's why (for cases when A is to the left of the curves) they called (in background) the rule of B as taken directly B=A*E/2=f/2 assuming that you know E in this limited linear domain. I mean it appears that has been decided is useless to improve the graph as drawn starting in zero- stress and zero-strain with additional 5 strait lines (case in which no A will be to the "left" of the curves)

I think the above will clarify your doubts. For your case, the troubles are not due to temperature but to curves themself. For "A" value of 0.0034 you are in the non-linear domain stress-strain and you cannot compare the value with those given by "step 7" which is intended only for linear domain not shown in figure CS-2 (also theoretically in agreement with the linear domain shown in CS-2).