The difference is that it's in the appstore. That's the difference, and that will always be the difference. You can't start differentiating apps in this way, it will never end and then there will never be clear rules. The appstore has guidelines, safari does not.
But an alternative browser could be in the AppStore. Personally I use Brave, a browser with lots of privacy conscious features. It's sending my inputs to an arbitrary remote server, which is then executing logic based upon that. Some of these things are transactions. Why doesn't Apple get a 30% cut of that?
I'm not trying to be an asshole, but I genuinely don't see the difference between these things. Both a browser and xCloud are programs which capture your input, send that input to a remote server, then receive instructions from the remote server on what to render locally on the phone. The difference is that with a browser it's uploading form-data and JSON and downloading HTML and JSON from an infinite number of hosting entities, and with xCloud it would upload a raw octet-stream and download a video from only a single hosting entity (Microsoft).
Actually typing it out, I think this is the key difference; it's impossible to be a middle-man for browser transactions because browsers cannot detect that it's happening, and there are an infinite number of hosts so it is not a tractable problem to partner with all of them. But with xCloud, Microsoft is the only host and can detect when a transaction is happening, so it is tractable for them to do it. I don't think that makes it right, but that is why it is happening.