> If the language is using an exclusive upper bound, then it is correcting the mistake of the programmer for specifying a length beyond what the string contains.

No, it's not-correcting the not-error of the programmer specifying the correct upper bound using the half-open-interval model, a common mathematical model of range specification that has lots of useful properties.