Over-engineering indeed, but it does address actual problems.
Primarily: Addressing what most browsers haven't well managed to handle: interaction with big files.
E.g. huge database-like Plain Text: browsers typically fail with OOM (Out of Memory).
I'll write a review of GitHub's delivery infrastructure when I have the time.That's the heart of it, isn't it? It's challenging.
At the core of modern web development is the attitude that developer convenience trumps all.
The software is free after all. (The thousands you pay in bandwidth and hardware upgrade costs to keep running their bloatware doesn't count because none of that goes to the web developer, and is therefore irrelevant in their it's-all-about-me worldview.)
Slightly off-topic: I've been viewing YouTube on a old YouTube frontend that works back to (at least iirc) IE6 and Win98 for some time now [1] [2] [3]. The frontend feels mad snappy as hell and it loads fast compared to YouTube today since the frontend is not heavily JavaScript reliant, it just uses it to enhance the site.
[1] https://i.ibb.co/YpmFP91/yt2009-34.png [2] https://i.ibb.co/THK0wk0/yt2009-35.png [3] https://i.ibb.co/nk7nxZQ/yt2009-36.png
Abuse of animation and skip of best-effort loading:
This is probably more relevant to the sluggishness and bad interoperability, rather than JavaScript.
[ Those abusing JavaScript for nefarious purposes: probably not the JavaScript to blame. ]
Standards established don't necessarily imply being well-designed:
Blindly following without thinking shall regardless trap.
Many current infrastructures are fundamentally flawed and difficult to fix:
1 step wrong, all steps wrong.