There's nothing wrong with the definition you quoted. I would say it's entirely accurate (except perhaps that it neglects subtitles, which can end up being a substantial amount of work if you choose to put the effort into it).

>What the author is referring to would be called a “hybrid remux”.

No. While the "hybrid" term has been used this way, the standard definition now is for remuxes where a single stream has been created by stitching together streams from two different sources. For example, a video from source A has generally better picture quality, but has some glitch or issue that requires patching in a segment from source B to create a hybrid video stream. Doing this for the main video without re-encoding is very difficult/unlikely to work because video codecs are very hard to work with in this way (you'd have to have two sources encoded with the same codec, the same encoding settings, you have to slice things at a clean GOP boundary in both sources, etc), so most commonly hybrid remuxes refer to the Dolby Vision layer.

Source: I happen to be involved in this community.