Rate an AI agent reply as the signed-in operator

Records how the signed-in operator rated one AI agent reply, on the same three faces the visitor rates replies with, and optionally why. Each operator holds one rating per message: rating again replaces theirs and leaves their colleagues' ratings alone. A comment explains an unhappy or neutral rating; leaving it out keeps the comment the operator already wrote, an empty string removes it, and a happy rating never keeps one. Operator ratings are team-only: no visitor read, visitor event or webhook carries them. Emits MESSAGE_UPDATED_EVENT to the dashboard.