RMS is nearly always associated with power, but there are many times you want to know power that don't involve heating resistors. For example:
The power sent to a speaker becomes sound waves. The RMS value of the waveform corresponds to the perceived loudness of the sound.
Signal-to-noise ratios are power ratios. Therefore, you want to compute the RMS values of the signal waveform and the noise to get a SNR value.
In electrical power transmission, it's the RMS values of voltage and current that correspond to the power delivered to, say, a motor (assuming the power factor is near unity, of course).
There are many more examples. The point is, in most forms of AC signal analysis, the RMS value is nearly always the more useful value — although there are times when you need to be aware of the peak value as well.
This mostly depends on what your powering the resistance wire with. The voltage going in, combined with the resistance of the wire will dictate the current flow and therefore the power dissipated in the resistance wire. All power dissipated in the resistance wire is useful heat, to heat up your materials (chemicals/shisha etc).
So the actual resistance is not relevant if you have absolute control over the voltage being fed in. As a rough indicator, a 10W heater with a 12V voltage source would need a total resistance of 14.4Ohm (Pd = V^2 / R), which with this wire would need to be 3.42m long.
How much thermal power you produce (Pd) will then dictate the temperature, depending on the materials you are heating. Thermal energy will be needed to bring up the materials to the desired temperature, and less energy will be required to keep the materials at the required temperature (due to heat escaping to the environment).
To keep a constant temperature, you need to reach a state of equilibrium where thermal energy added to the system matches thermal energy being lost. Heat loss will increase as the system temperature increases, so for any given heater system it will naturally reach a maximum temperature. However, any change to the system will change the maximum temperature (remove materials, the wire may overheat).
In short, you can make a heater system that does not have a controller, but the required power fed in is highly dependent on the system (what materials you are heating, how much etc). Your best bet is to use a lab power supply, in current control mode, and slower creep up the current until it heats up and stabilizes at the required temperature. You can do calculations to estimate heat loss, but they may not be much more useful than empirical testing.
As a rough guide for somewhere to start, make an educated guess as to the rough order of magnitude of power needed. This device heats about 0.5sq m of soil to around 30C, and is rated at 27W. So we could infer that for temperatures of 30-50C, over an area of about an A4 piece of paper to 0.5sq m, 10-30W would do the trick. This is assuming that the wire is heating something else up first (some medium like sand). This is a Fermi style estimation, to give a rough idea or scales. If you start testing with the ability to generate 10-30W of heat from your wire, you can then ramp up your power supply across the range to see what works.
If you do make a heater without a controller, you still need some kind of safety trip to stop it overheating. If you do some googling, you can find various ways to do this. The easiest is probably a simple thermal fuse in series with the resistance wire, that will disconnect power to the resistance wire if it heats beyond a set temperature.
Best Answer
Your calculations are slightly off, as the duration you averaged over was 3 seconds, not (2 + 0.5):
(2 * 2 + 20 * 0.5) / (2 + 0.5)
average current draw= 5.6 mA
2400 / 5.6 = ~ 428.6 hours
running time= 17.86 days
This is a good enough method of estimating run time.
Actual results will vary due to battery condition, charge status, temperature changes, and of course whether your current measurements are reliable and repeatable.