Hello,
As it happens, I just read about this issue. Briefly, the influences you mention tend to get balanced out by other influences, and they are accounted for in the conservative assumptions, the SIF, and the safety factor of the code design.
Peng's _Pipe Stress Engineering_ 2009 pp46&69 addresses this issue.
One of the conservative assumptions the code uses is that the hoop stress is calculated based on the outside radius of the pipe, even though the pressure is applied on the inner radius, then the stress is nonuniform through the thickness. This assumption eats up many smaller effects.
Also, Code SIFs (used in the EXP case) are conservative and based on real tests using real pipe, so the thinner wall property is already accounted for. For the sustained case, I understand that using code SIFs or other SIFs tend to take this and other factors into account (but some people don't use SIFs in the SUS case...)
Lastly, and most importantly, there is an offsetting principle called Lorenz factors. If you look at the section of a pipe bend, you'll see that on the intrados, the internal face experiencing pressure is larger than the external face (think trapezoid), while on the extrados, the effect is reversed. That means that when compared to a straight pipe, for each region exposed to pressure, the intrados has less metal area to resist the load, while the extrados has more.
So for a bend of even thickness, the stresses on the intrados are slightly higher than the extrados... which means that in a real-life forged or rolled bend, the thinning on the extrados tends to balance with the Lorenz factors.