For the most part, the TIA codes (or at least the F and G standards) tend to think of monopoles failing only by two methods. Bending and local buckling.
You can see how this would be the case for compression forces because all buckling failures are really just bending failures. And, if you're taking the 2nd order effects into account during your analysis, then axial forces effect on bending failure is already considered.
For this reason, I think that localized tension yielding of the extreme fibers would be allowed. After all, how are you going to achieve a maximum bending moment of Mp if you don't allow yielding at the extreme fiber. However, you do want to make sure that you avoid fracture. So, that might be a consideration for your connections.