They cite a paper by Kim et al. that calculates the thermodynamic limits on the specific yield of atmospheric water harvesting as 5–50 ℓ/kWh (0.02–0.2 kWh/ℓ), which I suppose depends on the air temperature and humidity.
The terminology you need to google for "the heat that is created when the water in the air turns into a liquid" is "enthalpy of vaporization of water"; around room temperature this is about 2.4 MJ/kg, so 10 liters/day is 24 MJ/day, which is 6.8 kWh/day, or, in SI units, 280 watts. However, remember that a heat engine operating at the Carnot limit is reversible. The coefficient of performance of a typical heat pump at these temperatures, the kind you might buy off the shelf at a big-box hardware store, is about 2, so you only need about 12 MJ/day (3.4 kWh/day, 140 watts). At a typical desert capacity factor of 25% this means you need a 560-watt solar array, about US$120 and three square meters. (California's utility-scale PV average capacity factor was 29% last I looked.) Very cloudy and polar places can have PV capacity factors as low as 10%, but they also have easier sources of drinking water. Like a rain barrel.
It's easier to store water or to "store coldness" than to store electricity, so you don't need electrical storage, you just need a heat pump sized for your peak throughput instead of your average throughput. Heat pumps are pretty expensive, so you might think this is a big problem, but the cheapest air conditioners I can find for sale around here are about 2000 watts, not 500 watts.
I've been noodling on desiccant-powered heat pumps for this and other uses, which may be able to reduce the cost of such systems, gather a larger fraction of solar energy than the 21% of high-efficiency PV panels, and provide built-in energy storage.