It really doesn't. Let's ignore that most of those tanks are in active use and can't just be filled and emptied at will so you'd need a lot of additional hardware to do anything.
The fundamental issue of gravity energy storage is that it needs a lot of height or a lot of mass (ideally both): 1kg lifted 1 meter is a bit under 10j, which is nothing (it's 2-3 seconds of a led bulb). If you have a 5000L water tank on the ground an an other on the roof of a two-story house, let's say 7 meters between the two, that's a bit under 350000J, or 97Wh, which is a laptop's battery. And that's before accounting for inefficiencies in the pumps and generators, and friction losses in the ducts.
A good PHS site has an upper reservoir capacity of millions (usually tens thereof) of cubic meters, and at least 100m hydraulic head (some are close to 1000).
You'd need absolutely humongous water towers (we're talking taller than the tallest and larger than the largest) to get decent storage in distributed completely artificial contexts: the tala tank (https://en.wikipedia.org/wiki/Tala_tank) has 1.5MWh capacity. That's a dozen electric SUVs (à la Rivian), about 20 sedans if their batteries are not too large.
Moving, lifting, and storing water are things we're pretty good at, humanity has been doing that for literally thousands of years, if it was an easy way to store energy we'd have been doing it, a lot, for a longtime.
Is it really surprising? Pumped hydro is essentially massive batteries with fairly high rates of in and out flow, so they're really good both to smooth out production variability from renewables and to match / lead rates of consumption to relatively slow-following nukes.
The problem of pumped hydro is having suitable sites, I believe the US is still the only country to have built a completely artificial PHS facility with Taum Sauk.