Old post, but the question keeps coming up. This is the only place I've seen Ernie address safety issues directly.
We usually shy away from this discussion in permies, as there are a lot of ways to have things go horribly wrong.
There's also a nice video from one of Geoff Lawton's experiments here:
https://permies.com/t/25971/rocket-stoves/Ernie-share-thoughts-Geoff-Lawtons#204995
As a general note: This is expert territory - meaning training and experience.
I've done minor plumbing (replace faucets, stop a leaking toilet) and have a degree in physics. My favorite way to heat water on a rocket stove is a large kettle or pot. Although I hope to learn more eventually (and have really enjoyed online research and working with experts like Caleb and Ernie), at present I would much rather take sponge-baths than try to build my own water heater. My grandmother did just fine with bucket-baths both in youth and in old age. It's safer for many reasons.
It's not that hard to heat a large bathtub, either, if you are willing to haul the fuel. You can do some quite clever immersion heaters or batch-heated water (like an ancient laundry cauldron), without needing to re-invent indoor plumbing in an off-grid situation without reliable power for pumps.
If you have never done water heating with natural fuels, consider that the heat output based on different types of firewood can fluctuate by a factor of 4. You do not want a "super-efficient" piped heat capture system, because overheating and steam explosions are your biggest risk factor. Big tanks like a stockpot over the stove, a Mexican hot water heater, or the
Lawton demonstration system are good capacity buffers so you have some time to stop the fire before it overheats. (Gotta learn the name of the guy who maintains Geoff's system! I think it was a collaboration, or Geoff is showing off someone else's design on his site)
I would suggest that most people are better off starting their hot-water-design careers with steps like installing a factory-made system, or building a kit with well-known design rules and a predictable max. power input (like solar hot water or refined fuels). Then work up to the point where you can anticipate how much water you'd need to buffer a variable-input system like combustion fuels.
-Erica W