The title is misleading. They've dropped the proposal as applied to Chrome, but are still pursuing it for the Android WebView API, which is basically a wrapper around Chrome.
Is there a reasonable angle to view this from?
I personally don't think embedded webviews should be allowed general browsing capability unless they are part of a standalone browser.
It's usually a trick to capture traffic that would otherwise go off to the open web.
Webviews to the appmakers server need to be authorized by some manifest file on the server whitelisting the app identifier.
Webviews for the wider web don't allow the app to know what's going on inside the webview, nor interact with it. So these are safe to type passwords into etc.
How would you propose doing this from a technical standpoint?
How is android supposed to know what an embedded login is on a web page?
What happens when you're in an SSO or other secure environment and need to refresh credentials after a redirect?
That's a difficult problem to solve though. To do it properly you'd need something like Windows' secure key sequence (ctrl-alt-del) which apps can't intercept. Otherwise there's no way that a user can know that what they're seeing is the system rather than a malicious app.
Consider that any app can embed any browser or UI that they want.
foo.com could also cooperate and allow the message to say:
"Appname wants to 'send pokes' on foo.com, do you want to allow this?" - allowing the app a scoped login to only specific actions.
Of course a malicious app could still just show a webview and make you type the password in. The above solution would only work if everyone uses a password manager and showing a webview becomes way more suspicious than it is now.