Handling timeouts and refetching failed resources is one job for the browser, which are quite good at it nowadays. There is no need to badly replicate (with extra bugs) this tedious logic in Javascript for each and every website.

You can, but it's a lot harder than building a very fast website that just uses Express to render Pug templates and that caches HTML in a CDN.

If the site is static enough to do that, why wouldn't I just use a static site generator? Not being snarky, I'm really trying to learn.

You might need to retrieve your document content from a database. The CDN can phone home every so often to check its caches are still valid. This is basically how we served The Guardian.