Re-inventing the wheel
So, having determined that it is "obviously" better for everybody's apps to run on Google's CPU, they are now laboriously trying to re-implement the Windows API on HTML. Today's contribution appears to be some kind of registry where applications can be marked as "handling" particular content types and other apps can let the user pick from the available handlers.
Meanwhile, over the road, Microsoft have just finished re-implementing the entire Windows API on .NET, only to discover that no-one is really interested because running everything on an emulated architecture is *so* last decade compared with running everything on a VM in a random data centre.
Meanwhile, round the corner, the WINE people have nearly finished re-implementing the entire Windows API on Linux, only to discover that actually the documented API isn't the set of goalposts they should have been aiming at and so consequently only half the apps work.
Meanwhile meanwhile, anyone who actually wants to get stuff done is running the original Windows API on the original native hardware and wondering why the industry hasn't done anything innovative in the last decade or more.