Add segment-level confidence flag

Provide an optional flag for each segment indicating the model’s or system’s confidence in the suggested translation. Helps editors quickly prioritize segments that may require extra attention, without forcing interaction for every segment. Can be toggled or ignored by the user.