I did a quick test and have continued my research, and calculated that we could save a minimum of 1,500 gallons a year. That's with very conservative values, and assuming that we hop in the shower as soon as the water gets hot. With the tankless heater and the new copper plumbing, it took less than 2 minutes for the water to get hot for a shower.
In reality, I don't stand there with my hand in the shower waiting for the exact moment the water gets hot, and I think our shower puts out more than 1.5 GPM. A more realistic estimate is 5,000 gallons a year. With current water EBMUD rates, that's about $16 a year in savings.
Which means it would take anywhere 12 to 82 years for a recirc system to pay for itself. That's not counting the small amount of electricity it would use on the minus side, and the unknown amount it would save on gas for heating water that's wasted on the plus side. So there's not much of a financial incentive to do it.
But I'll probably still go ahead with this, because while 1,500 to 5,000 gallons doesn't sound like much compared with annual usage totals, it's the equivalent of anywhere between 900 and 5,500 flushes of the toilet in a year. That's a lot of perfectly good drinking water to just flush away.
On the other hand, while rain barrels are a very cool, low-tech solution for saving water, even if I had nine 55-gallon barrels under the various downspouts and linked to each other, that's only 500 gallons that I could collect at a time. Given the wet winter and dry summer pattern here, that means besides some condensing fog, the last 500 gallons collected in April and May wouldn't see any company until October or November. So that may require some more thought, unless I get some free food-grade barrels offered to me.