What a nice question. My own intuitions used to lead me up the wrong path here. When I was young, I had toy boats (now I have big boats, but don't call them toys.) The first had a spring to drive it and when I lifted it out of the water, naturally I stopped the prop to prevent the spring from unwinding. When I got an electric one, I followed the same practice, imagining I suppose, that by stopping the prop I was saving my battery. (It was easier to stop the prop before I reached the switch and disconnected the battery.) Of course I know now, that the better course is to let the prop run free until I can switch off. A free running motor draws less current than a stalled motor (by a long way.)
Your issue hinges on this constant voltage supply and P=VI. If V is constant, then P∝I. So if no current flows, no power is used - the device is switched off. When it is switched on, more current = more power. A light bulb has perhaps a 3A fuse (generous), but a 3kW heater needs a 13A fuse.
So the light bulb must have a higher resistance and the heater a lower resistance to allow a suitable current to flow.
Another formula you get from ##P=IV and R= \frac {V} {I} is P = \frac {V^2}{R}##
So power is inversely proportional to resistance for a given voltage.
Finally, there is the maximum power transfer theorem, which can be easily derived from the above, that says the maximum power you can draw from a supply occurs when the load resistance equals the supply resistance. If the load resistance is higher or lower, then you get less power out.
The extremes are infinite load resistance with no current flowing = no power out (obvious again) and zero load resistance or short circuit with large current flowing. Not quite so obvious that no power is dissipated in the load, because in practice we don't get absolute zero resistance shorts. But the supply (eg. a battery) gets very hot because that is where nearly all the power is being dissipated.
The mains power supply has to have a very, very low resistance, otherwise as soon as someone started drawing current, the supply voltage would be reduced and everyone else would get less power. So any load you connect is going to have a higher resistance - above the resistance of the supply - and reducing your load resistance brings it closer to that supply resistance, allowing it to draw more power.
So much for the math /physics argument, which maybe convincing, but may not be satisfying intuitively. Let's get on with the arm waving and hope the real scientists don't get too upset.
Say I want to push something along a flat surface. If it is very heavy, there will be lots of resistance (friction) and I may not be able to move it. I will push hard, work up a sweat, dissipate lots of heat inside me, but the thing doesn't move, I'm doing no work on it, it doesn't get hot from the firction of the ground. OTOH if it is very light and the surface is smooth, I'll move it easily. If I do as before and push as hard as I can, I end up running flat out, again getting hot, but still not doing much work on the object, which also isn't getting hot from the low friction. Somewhere in between are things I can push with moderate effort against moderate resistance. Both I and the object will get warm, me from the effort and it from the frictional heat derived from the work I'm doing on it.
We are rather complex power sources, so I can't say the rules are as simple as with electricity,but you can imagine there is an optimum load, where you can do the most amount of useful work. Deviate from that and the useful work done must decrease.
A dam full of water has lots of potential energy. If we just open the valve and let water pour out, it will rush out fast, lots of turbulence and that energy ends up as heat in the water as it slows down in the lower pool. If you put a waterwheel (for intuitive simplicity) in that flow, the water will be slowed down by it, the wheel will experience torque, rotate and useful work can be extracted. Less energy wasted in turbulent heating of the water. A light wheel (just so that we can ignore efficiency) with a light load will whiz round easily and extract a little useful work. The water will still flow fast and generate turbulent heat. Increase the load and we slow the water more, get more work out and less waste turbulence. You reach a load where the pressure of the water is turning the wheel at a good speed, generating lots of work, but if you increase the load any more, the pressure is insufficient to turn the wheel faster and it starts to slow down. Either the flow is restricted and now less energy is coming out of the dam, or the water can flow past, so more and more is flowing past without doing useful work and more energy is again ending up in turbulent waste. In the end, if the load is too great the wheel stops and no work is done. Again, making the resistance of the wheel closer to the optimum for that supply enables it to extract more power.
(Somewhere on the web was a beautiful video of a Pelton wheel being adjusted to optimum load. At low loads or at high loads water splashes off at high speed forwards or backwards, but at the optimum speed it almost dribbles away sideways, its energy largely spent. Unfortunately I can't find it at the moment.)
Hope that might be some help. I'll mull it over and try to think of better explanations if I can.