Once SafetyNet was brought to Android a decade ago the tendency has been clear - these freedoms are going to be restricted heavily.

Because how do you make sure it's the user who does those modifications, willingly and well-informed? That it's not a malicious actor, not an user getting socially engineered or phished? Incredibly difficult compared to the current alternative.

If it's not a software root of trust that provides an attestable environment like Android or iOS. It's going to be a hardware root of trust that provides an attestable hardware environment, like SGX. I can predict no other practical avenue taken. Unless the orangutan really forces a demonstration on how untrustworthy these environments can be and a lot of money and effort is spent.