At first I thought you had a point (and for practical purposes, perhaps you do) but I don't think it'd be particularly challenging to inject video frames into a stream without re-encoding the original video.
It’s not that difficult a command compared to what is possible with ffmpeg. But it’s also a pretty common enough task that literally the first result for using DDG with “ffmpeg concatenate without reencoding” gave: https://stackoverflow.com/questions/49371422/how-to-merge-tw...
(Edit: my children comments are right)