I think the "new protocol" example is particularly good one - things like IPFS run complex daemons under the hood and might access gigabytes of data, while their interface to browser is relatively simple: a channel which takes small requests and returns medium-sized responses. It's ideal case for moving to native code.
(Of course it has its own downsides too, like losing ability to run same codebase on the web. So far, I've been avoiding Electron's usage by running web versions instead - Slack-on-the-web is just as featureful as Slack-in-Electron, but shares memory with my main browser. But for other applications, native code should be a plus)
This whole article is about Chromium and Electron. I am not sure what the purpose of your question is.
I get why people are upset at that (especially the sometimes significant performance issues), but I also get why many applications went down the "Electron everything" route. In the end people who know C++ well enough to be able to deal with cross-platform bullshit are expensive and coding for cross-platform is a nightmare. There's a reason why there aren't many "top dogs" left.