You're making this way harder than it is.
AC electrical wires throw a considerable magnetic field
in proportion to the current on the wire.
And there's no particular need to detect the amount of current on the wire - 10A vs 20A vs 30A - because resistive heating elements only draw current at one of three rates:
- 0A
- design A
- infinity A, however this does not need to be disambiguated; the circuit breaker will trip and change this to a 0A reading.
Thus we are down to "current or not" as the only thing we need to detect. That's pretty easy. We can do that with magnetic reed switches, or better, Hall Effect sensors.
It simply boils down to finding the right physical places to surveil the wires.
Anywhere along the green for the upper element. Anywhere along the red for the lower element.
Anytime you're trying to sense current, you must remember the golden rule of AC mains wiring: NEC 300.3 all related wires must be kept together in the same cable or conduit, so that currents are equal/opposite in any group of wires, and thus, the magnetic fields thrown by those wires cancel each other out. That is how we avoid eddy current heating, disturbing animals and attuned people, vibration leading to metal fatigue, and other harms. However, if we are trying to detect current, this rule is our enemy.
So we must carefully choose locations where the wire under observation is not next to other wires. Or alternately, twist them like this so the magnetic fields stack.
Do not intermix AC mains and low voltage wires
A cardinal rule of AC wiring is that you do not put conductive low voltage wires inside the same enclosure with AC mains wires, and then, have low voltage wires leave the enclosure and go somewhere not protected as well as AC mains wires. So with low voltage you have to create an optical or radio isolation - either keep ALL the low voltage stuff inside the enclosure including its power supply, or keep it all outside the enclosure and use magnetic fields, radio or optics to communicate through the membrane.
Water heaters typically have several access covers. If it were me, I would play with positioning the relevant wires against the access covers to see if the magnetic field will punch through the metal enough to be detected by a Hall Effect sensor (a reed switch is probably too much to hope for lol). Then your installation is neat as a button: all the low voltage DC stuff is on the outside of the steel enclosure and you have achieved separation. Otherwise you might put standard 1/2" knockout holes in the enclosure and filling them with approved plastic knockout covers (generally intended for a Romex cable, but I won't tell lol). One thing or the other should do the trick.
Then it's just Arduinos and Hall Effect sensors, straightforward stuff, the folks over on electronics.se can help with implementation when you get stuck. It's up to you whether to stick a WiFi shield on the Arduino so it can send you email lol.
You may notice that knowing "when the element is on" is slightly different than knowing "when it is supposed to be on". I think you can handle that in software. Just keep track of the last time the element has been seen in use, and alarm if it's been too long.
