When you ‘stream a video’ from firefox it just downloads the video in small chunks at a time instead of the whole thing at once. These chunks of downloaded video are saved to temporary memory called a ‘cache’ and deleted after you are done the video.

Yes yt-dlp is most often used to download the entire video as a digital file onto permanent memory; however it doesn’t have to be used that way. Other applications like smplayer and mpv can work with yt-dlp. Using it as a component to do the heavy lifting of talking to youtubes servers and streaming video in the same exact way firefox does.

Doing a quick search, there are some projects to implement mpv with sponsorblock. Im not the most technical person and prefer not to get my hands dirty with complex hacked together scripts that require compilation or whatever. Thats not to discourage you if you want to follow up on those things know people are working on it but if you aren’t a power user it may be a hard time to get that kind of thing working.