As in post #4, the equipment is called a “cold trap”. You can Google that term and see many for sale for applications very different than this one.
The actual physical limitation in your scenario would be heat flow into the surface region of the water. That surface will cool quickly, as every molecule of water you pull away costs you one chunk of vaporization heat. That heat has to come from somewhere: convection is your best bet, limited by practical choices on the ground.
If you pump too fast, you’ll freeze the surface. So, you’d need to throttle the pump to draw just below the rate where it would overshoot the input heat. This rate will vary with the environmental conditions.
If you use a fat pipe and if (mostly) convection can get tens of watts of heat flowing continuously to the surface, and you don’t accidentally freeze the surface, you could maybe get a gallon of water every three days when the lake is warm.