If memory serves, the upper temps in B31.1 are not mandatory limitations. Though concerns for transformation is real and should be considered. Knowledge of this alloy is critical to successful welding.
As for fillers, the current popular recommendation is B9 (though I have been out of the Grade 91 loop awhile), which is perfectly acceptable, but consider that when using a B9 you will, with PWHT, have carbon migration from the P22 HAZ into the B9 (carbon migrates to the higher alloy side of the fusion zone) weld metal weakening the P22 HAZ. This loss of carbon being primarily responsible for the infamous Type IV cracking. Whereas when using a B3 filler the carbon migration will be from the B3 filler to the P91 HAZ. And even though the B3 weld metal will be weaker than a B3 used for a similar P22 to P22 weld (where carbon migration doesn't exist), the B3 weld metal will still be stronger than the P22 base metal.
Therefore, the strength gradient using a B3 filler is more linear across the entire weldment with B3.
And the other issue is, where would you rather have the high carbon zone? In the BM HAZ or the weld metal? With B9 the high carbon zone is in the B9 WM. With B3 it is in the P91 BM HAZ.
Add to this the fact that B3 is not as crack sensitive or heat regime sensitive as B9, and is cheaper, my opinion is that B3 is acceptable for this dissimilar application as well.