All the reasons the author lists is primarily why I've been preferring using web components to add interactivity to web pages only when the components need them. Nice thing about web components is that it's browser-native so it doesn't require any framework code to ship with it, and if you do prefer a framework, most modern frameworks have wrappers that work with web components.

We ended up building (yet another) framework dedicated to rendering web components (https://rootjs.dev). It uses JSX for rendering HTML on the server-side, but no JS code is rendered to the browser at all by default. Web components usage is auto-detected and automatically injected into the page as a dependency, which makes for really good developer ergonomics.