I hadn’t heard of the SCS bus before. Using an entire separate microcontroller for bit banging it seems like a pretty sensible straightforward way to avoid having to coax hard-real-time behaviour out of the ESP32 chip that is also in the background taking care of various complicated things like wifi, TLS and MQTT. I wonder if you could get even lower power consumption by using a really tiny CPLD or FPGA for that?
After reading both this and the linked previous blog post, I still kinda wonder why the Nuki device? Even with the reasons given here. It sounds like that device is actively hostile to the entire concept of doing things that work!? Such a large portion of the technical difficulties described were faults with it.