Thanks for the recordings and the pictures, that explains a few things.
What I meant by “without the 100 Ohm resistor” was basically open-circuit to see the zero reading, something that I could see in the 12:53:00 and 12:58:27 recordings, which shows that this is not an offset issue.
When I’m looking at the breadboard wiring I believe you may have additional resistance from the breadboard to the shunt resistor. Your multimeter is clipped on fairly close to the resistor body, it would be much better if the ADC+/- pins could be attached similar to that, right next to the resistor body and not through the breadboard contacts. Contact resistance can easily be in the tens of mΩ. I believe that explains why the readings are off (the shunt voltage gets measured on the “wrong” side of the breadboard contacts which carry the current).