The problem with the temperature theory is that liquids don’t tend to have much of a thermal expansion coefficient. A quick google search turned up 950x10^-6 cubic ANGSTROMS per degree C. AKA nothing.
As for pumping more slowly… Again liquids aren’t compressible so having the nozzle half open will result with a higher volume but at the same velocity so I don’t really see a lot extra being vaporized. Maybe from turbulence in the tank lowering surface tension, but still.
Just theorizing.