This is an interesting question and really goes to the root of a major difference between ASME and AWS. IMO, ASME takes the idea of tensiles and bends literally, I suppose is a way to put it. In other words, you are verifying mechanical viability. Strength and ductility. It is assumed that the material will be free of flaws. Flaws are implicity segregated to welder ability. In other words, if your welder is putting slag into your PQR coupon shame on you.
AWS don't see it that way and mixes in performance variables with procedure variables.
Now, IMO, its not quite as clean as either approach would emphasize, since running too low of amps could clearly cause slag no matter how good your welder is. But even though AWS has some justification, which may be related to a concern for prequal thinking, (just guessing here) I do think the ASME approach is closer to reality. Welder incompetence should not be an issue for procedure qualification.
So in your case, think aobut it, evne with bends some material is coming off. you don't want to bend a rough cut. So, IMO, its a question of how much your removing. And how deep your flaws are.