Quick question: is the embed structure wrong on purpose?

what do you mean?

I'm most likely missing something but I would have expected the embedded video to be within embed -> video (just like images in the other example). Instead right now all properties of the video struct are just one level up right within embed.

at a guess, I think images are like that because it’s an array

I thought I saw the same for record. Let me find an example

Here is an external embed that is also nested within embed