<here comes the problematic part - do not hate please >
Call me stupid but the way of my thinking is: if battery would be pushing constant current through resistor (by increasing voltage or other method if exists) then battery life would be always the same, ignoring the value of resistance. It's like battery would say: "i'm giving 1A no matter what - I want it to be over in 30 minutes" - this would lead to putting the resistor on fire. But it does not happen. It means that resistor is in fact stopping battery from pumping the whole current it could deliver. It limits the current flow. But then why Heat? So the question is: HOW resistor is limiting the current?
Clearly, you describe something which is manifestly
not a battery -- i.e., something with ~constant voltage output, and obeying a charge-current relationship.
Instead, the constant-current analogy to a battery (which as it happens, doesn't seem to exist, outside of contrivances, whereas the battery is a simple chemical process!), would be a ~constant current output, obeying a flux-voltage relationship.
An inductor would be the linear example, just as the capacitor is the linear version of a battery. By "linear", I mean to note that the charge-voltage or flux-current relationship is proportional, i.e., voltage increases proportionally with charge, current increases proportionally with flux.
A battery is NOT a linear component (so, if we model one as a Thevenin voltage source (an ideal voltage source plus a resistance), we must observe that its equivalent resistance is
non-ohmic -- aha, bringing the off-topic discussion back into it, see?
), so our analogous component must also be nonlinear. In particular, it needs to have ~constant current over most of its charge, until it becomes depleted and charge goes to zero.
Real (ferromagnetic cored) inductors exhibit saturation, but this is the opposite effect: as flux goes up, the rate of flux per amp (the inductance) drops, so as you continue to charge it, the current rises exponentially, rather than leveling off.
You could make a locally battery-like inductor, by pre-saturating the core with a permanent magnet. As flux builds up, the magnet is opposed, and inductance increases. Downside: this doesn't continue forever; once you go over the center hump, inductance goes back down again, mirroring its initial rise, merely offset.
If moving parts are allowed*, a solenoid may be a better example. As magnetic field rises, magnetic force tugs on the armature, opposing the force of a spring. As the armature pulls in, the magnetic path is shortened, increasing inductance -- ah ha! As the path shrinks, more and more flux must be added to increment the current, i.e., inductance rises more and more. Eventually, the solenoid is fully charged (armature fully seated) and further charging only saturates the core again. Downside, most solenoids actually pull stronger and stronger, as they close, due to geometry; you may find a real solenoid actually has so much inductance in the active region, it's actually beyond infinite and becomes negative -- which is another way to say, it exhibits hysteresis (the solenoid tends to stay pulled in, until the (relatively low) holding current is removed). A solenoid could be shaped to have a flat flux-current curve, though.
*A battery has moving parts: charged atoms (ions) moving between electrodes, through an electrolyte. They're just invisibly small...
The downside to a magnetic component is, the time constant over which it stores useful energy is very limited. Current flow through the coil causes resistive losses, effectively giving a "shelf life" of milliseconds. Conversely, eddy current losses in the core (when the magnetic field is changing, or when there is relative motion) make rapid dis/charging extremely inefficient, and the mass of the armature itself limits how fast mechanical motion can be transformed into electrical energy (again on the order of ~ms).
A solenoid made with superconductors would be pretty good, though.
Incidentally, there are actually superconducting energy storage devices; a coil akin to an MRI magnet (just smaller; more like a chemical NMR machine, microwave-oven-sized) is charged with a few thousand amperes, and some very beefy transistors keep the coil short-circuited most of the time, but allow that current to flow into a load as needed. Because the inductor is air-cored, it doesn't suffer from saturation, and because superconducting wire is quite fine, it can be made with a great many turns, making it possible to carry kiloamperes in a fractional-henry value inductor, with the only steady-state loss being the switching circuit (which itself can be made superconducting, if the delay from operating a mechanical switch is acceptable).
Incidentally, I think I once calculated that ITER uses a total around a kilohenry of superconducting coils, at a typical current of a few kiloamperes, for its various field coils around the meters-tall reactor; in total storing some gigajoules of energy! (They don't give you all the numbers, at least not without digging through design documents; but you can calculate these from some educated guesses and the numbers they do give.)
Oh look, I'm rambling about things; cool things, admittedly (get it, because we don't have room-temperature superconductors?..), but ah... guess I should go to bed.
Tim