The idea of someone deserving death in itself is deontological (some people must be punished and that's a rule) while talking about the net utility of whatever is consequentialist. Ethics should be impersonal (that is, treat everyone equally) so a consequentialist ethical system that doesn't approve of death in general should never approve of a death of any single person as an end in itself.

Generally, it seems to me that for a consequentialist, talking about an act or a person being evil should only be computational shortcuts over the real substance of moral reasoning (which consists of assigning utility to world-states). Like in the common example of an airplane that we describe using aerodynamics because that's convenient even though really it runs on the same fundamental laws as everything else. We tend to use those shortcuts reflexively without really thinking what we are trying to say in consequentialist terms.