The goal is to eventually delete all the desktop app specific code in Construct. A quick search shows over 60 places in the editor code where we branch off and do something different for desktop apps. It adds a lot of complexity and maintenance cost to the codebase, such as dozens of often awkward desktop-specific bugs, when we'd rather be spending our limited development resources on other things,. It also kind of defeats our generally very effective strategy of having a single product and single codebase, which is one of the reasons we've been able to keep up with much better resourced competitors.
Removing this code means even third-party launchers will no longer be able to activate desktop features in Construct 3. Besides we could not offer support for that code if anything went wrong with it after that part of the product had been officially retired. Therefore even if we just left it behind and ignored it, there's a high chance it will eventually break and stop working anyway. So I'd rather fix any remaining problems, then retire the desktop builds, delete all the code they used, and get everyone to migrate off it in one go. Having a lingering usage of unsupported desktop apps with people still getting upset when it breaks is something I specifically want to avoid. I think dealing with the pain of a change up-front is actually better for everyone in the long run.
There are still a few things to do, such as sorting out an NW.js preview alternative. I think Remote Preview should be able to handle that. Part of the plan is to identify these gaps and fill them in before we make the change. So please plan on the assumption you will be switching over - don't assume that there will be any support for desktop builds at all in the long term.