I feel like I might be falling into some kind of fallacy, but it seems at first glance like your argument supports GP's.
If Electron is such a horrible platform that you literally can't write performant software with it, and Slack's performance problems are really a huge problem for regular users, and yet as a massive company with a singular product they still rely on it, then I would guess at least one of two things should be true:
A) Native app development is actually way more horrible than most people say, and Electron's experience is so much better that even a company as big as Slack finds it preferable to building a dedicated Mac team.
OR, B) Slack's dev team just may not actually care all that much about performance in the first place, and they might just be writing a poorly optimized app on top of a poorly optimized platform which exasperates all of its problems.
Or, maybe a combination of the two. I dunno.
Anecdotally using Slack within a web browser (which should get rid of some of Electron's downsides since it's no longer a separate browser instance) I still occasionally see slowdown and issues with the App. Not nearly as bad as, say, Twitter, but not great. Certainly not as performant as I would expect an interface this simple to be.
So while it's silly to claim that Electron doesn't have an impact on performance, I suspect that even if the Slack team was writing native apps, they still might end up with a piece of software that has problems. Maybe fewer of them, or maybe even more, since you can get away with a lot of bad architecture in a C++ program before you hit the same bottlenecks like loading time or typing lag.
Are we sure that the Slack team wouldn't just view a native codebase as license to cut even more corners with performance?