As far as I know, of the many thousands of units worldwide, nobody has cooked the ADCs. The TVS diodes in the CTs are there to protect the CTs. Without them, if disconnected from the IoTaWatt while clamped to a current source, they can develop very high voltages internally and damage themselves.
The IoTaWatt has its own protection. First in line are the burden resistors. They are 20 ohm 1/8 Watt. 1/8 Watt would be 158mA input - about three times the 50mA design limit. Somewhere above that they would start to heat up.
At 158mA, they would develop about 3.16 volts rms. That would be +/- 4.5V. Sitting on the 1.65V bias that would be +6.15V/-2.85. There are rail-rail protection diodes on that signal that would limit those voltages to within the absolute maximums for the ADCs.
Which is probably why nobody seems to have had any problem despite a variety of mismatched CTs, both current and voltage type, having been deployed by adventurous users. The end result has always been just inaccurate measurements.