you're correct. Cross-file reps are segments that appear in several files. So once you've translated one file, the next becomes easier. It shouldn't matter in what order you do them. The cross-file reps appear in file 3, because that's the order in which the files are analyzed.
You can experiment - rename the files so they get a different (alphabetical) order and run the analysis again. Again the cross-file reps should appear in the last file that's analyzed.