I think the issue (as touched on in the article) is that devs don't want to make "browser games" so much as they want to make games. The main advantage of being able to render within a browser window would be to provide a quick playable demo. When you play an imersive game you want the whole screen because it is taking your full attention.

When it comes to games I'd prefer just to manage everything within Steam or via desktop shortcuts, whether it is written in C++ , C# or JS is an irrelevant implementation detail for the player.

Of course by providing an installer for popular OSs that just fires up chrome in fullscreen mode could easily solve this problem.

Another (possible) issue is that most game devs like to keep their source private and include DRM. JS doesn't lend itself so well to that type of approach.