If i understand correctly, as part 4 is set to loop, when the sequencer is stopped, all the tracks continue to run their audio. this allows you to launch another pattern (even from another bank) inaudibly in the background.
In order to mix the new track audio in however, the previous track audio, which is looping must be stopped and the new track audio launched. That’s 2 steps per track. One to stop. Another to launch.
By switching to part 1, where track loop is off by default, the looping audio from the previous pattern using part 4 will automatically stop looping when the new track is launched, thereby saving a step. I guess he decided that the extra step was a good trade off from having to keep switching between parts.