My sense is that existing physical access exploits have relied on legacy assumptions that shouldn't exist for wifi/bluetooth - hopefully no-one is building systems that give wifi devices direct memory access, or boot over bluetooth. They'll need some level of management interface, certainly, but that's likely to be something purpose-built and more tightly constrained.
Your 2 attack scenarios are: device is in evidence lock up, and can b cracked at will (so a security company probably comes up with a kit for cracking and hooking them up, sells the service to LE) - or what happened with T2, where the software stack just gets exploited.
Since you'll still need all those management interfaces that currently are only on the physical port to be somewhere - well I'd put money on removing the port actually making the device more vulnerable. Briefcase of GPUs to crack wifi and rootkit a phone if you can get on the same network.