So far target repetitions are calculated based on the contents of the segment at time of import (for bilingual files) and on the first entered translation (for files translated in translate5).

This was historically wanted but seems wrong today.

This should be corrected and what a target repetition is should be decided based on the latest content in the segment.