NW.js itself has been a regular source of difficult and complicated bugs and issues in the past, which is why I would prefer everyone moved over to using it in the browser instead. It tends to work better there.
The only reason now to use NW.js editor is really just for convenient testing of the NW.js plugin. But even then you could use the editor in the browser, and use Remote Preview for NW.js to test the NW.js parts. You could also design your events to still do something useful when NW.js is not supported, e.g. fall back to placeholder data instead of reading files from disk, so you could do most of your testing in-browser anyway and then just occasionally remote preview or even fully export to test the NW.js parts.
The new Windows and macOS desktop wrappers are much simpler and should be highly compatible with browser features. It's also possible to add extra runtime features like file system access in a way compatible with browsers (e.g. how Construct can use files on disk in Chrome). So in the long term I'd like to move to a world where the NW.js plugin is basically replaced by browser-compatible alternatives. Then there shouldn't be much need for NW.js left at all by the end - especially not for the editor. (But for backwards-compatibility, we'll likely keep supporting the NW.js export option for a long time anyway.)