After reading through the docs around IoT Hub quotas and throttling (https://learn.microsoft.com/en-us/azure/iot-hub/iot-hub-devguide-quotas-throttling), the throttling appears to be per the hub and not per the device, is that a correct assumption?
If that is the case is there an elegant way to set quotas per device so that particularly chatty devices can be throttled?
Yes, your interpretation is correct, Quotas and throttle limits are per hub or per unit of IoT hub. You would have to throttle it on the device IoT clients yourself. I dont think there is anyway to prevent IoT Hub from receiving a message from a registered and enabled device.
There is a provision to disable the device on IoT hub. You can explore if there is an API to do this programmatically and disable a device. And then you could enable it again after sometime. But this could be very problematic if not managed well. So think through your use-cases very carefully before you think of this option. IMO it is best to throttle from device itself.