Ingest not only applies IPTC and copies the images but renames the images and does some internal changes for example converting read-only bits to tags removing write protection. Of-course the first half can be done in two steps (one renaming, one applying IPTC), but it takes doubble time to set things and doubble time for PM to process all images.
If you use a limited amount of disk space on a laptop or a USB drive you already copied the images and there are coupple thousands of them it would be quite a different not having to wait for PM to rename all images then letting you aply the IPTC but after a busy day of work you can start it in one round not consuming all your disk space with copyint (and not needing an extra step of deleting all if you used ingesting).
I hope this makes my point somewhat clearer...