Again, I'm not too deep into WebSub, but I know the predecessor quite well.
This spec solves a specific problem. You have one URL resource many clients are interested in, that URL gets occasional updates, many clients want to fetch that new content. Think RSS feeds. so far, every client had to pull, meaning look at the file again and again and see whether there is a new update. Think a bit about that and you see how that is hard to do on client side (comparing content, making sure new stuff is really new and not just reordered, storing the old file, etc). Normally, those clients are servers. That's really important to understand, without that knowledge WebSub makes no sense.
Push-based protocols solve that problem with a middleman, the hub. Whenever there is an update the original server sends one single POST to the hub, the hub then sends that notification to all subscribers. And wush, no more polling.
The end result is way less traffic on the lines, way easier architecture on the feedreader side (it can be not built on a polling infrastructure, instead it just has one webhook open, and when it gets notified of an update it fetches the source once), less server load on the origin. For blogs for example that really is relevant.
This has nothing to do with a CDN.