"It sounds like if this UDP traffic was bridged onto your LAN, you could view or control
this via a phone app without anything leaving the LAN."
Homekit basically does this with multicast DNS. It removes a lot of latency between the device and the app and means that you can control the device when your internets are broked.
"but this setup isn't an automatic information grab at least."
I'm not sure why the reg started putting out this FUD but IoT companies don't want your data to resell it. They want the data so that it can be used to make the IoT product worth while. What would be the point of a "smart" device that couldn't tell the backend about what is going on? I think you're more likely to have your browsing habits from news sites like this resold than a bunch of boolean values that represent the state of something like a smart light bulb in your house to be honest.