That distinction matters because agents do not act for themselves. They act on behalf of someone or something else: a person, a team, an organization, or a system goal. Their authority should be bounded by that delegation, not by a broad identity-based role that persists beyond the scope and duration of the original delegation.
This para is where the penny dropped for me. Imagine a a general purpose agent that can do whatever you ask it to. It's meaningless to give that agent a role with fixed permissions outside of the context of a specific task request