We already have the frozen and boiling temperature of water or any other liquid.
I remind you that boiling water is 100c and not higher as the energy is used to change phase. (at 1atm and close to the sea level)
You can use different materials for the temperature you are more interested in if you want more precision, we don't need to go under bodies of water to calibrate thermometers
I think they meant that hypothetically, presuming a case where we didn’t have access to better methods. The question might be rephrased as “would this be accurate and reliable enough to calibrate a thermometer if someone cared to try?”. I might be wrong, but it’s how I interpreted it.
