The oil idea came from mineral oil passive / fanless cooled PCs, people has built them and they work. I don’t know if it can be done for flashlights, maybe it can’t, my knowledge about cooling is close to zero, but it isn’t just “add stuff”, oil has been used for electronics cooling.
Oil cooled PCs still require a radiator and cooling loop to cool down the oil.
Otherwise it would just heat up slowly until everything overheats.
Again, the heat dissipation is done by a radiator or cooling fins (in the case of a passive PC) on the exterior of the device.
Buy a big box of dry ice and you should be fine cooling everything you want for a while!
IMO water cooling is impracticable and too heavy. And a fan is noisy.
For EDC lights I prefer small size with efficient low modes, with short crazy high burst capability to see further down the road. No need for that to run indefinite. A hunting rifle can not be used as a machine gun, it will overheat. But who cares? It’s light and accurate.
For a long term search light, you obviously need more mass. I think every design has it own niche.
Mass is not what dissipates heat.
Surface area is.
Simply adding more mass will make it take longer to heat up and delay the step down.
It also increases weight a lot more than simply adding cooling fins, which is what actually cools the light.
For indefinite runtime you need good heat dissipation, not high heat capacity.
I have a better idea.
Get some liquid nitrogen (LN2), it is –196 degree Celsius or –320 degree Fahrenheit. 
The biggest advantage is that liquid nitrogen can flow trough a heatsink. You probably dont need a radiator to cool it down.
Disadvantage is that liquid nitrogen turns into gas at room temperature with a expansion rate of 1:694. That means it will create a huge force in a closed container… :person_facepalming:
The Lemax LX70-SP (HID) maintains its 5600Lm indefinitely. The secret? Its 4,3 Kg! (4,2 Km throw)
For the same reason the BLF-GT90 (2 Kg) maintains its turbo for 5 minutes whereas the turbo of the Acebeam K75, with about the same output, lasts just 90 secs.
The secrets are the correctly designed heatsink fins and that is uses an HID bulb which projects a lot of its heat out the front. These HID bulbs also tolerate heat better than LEDs.
The heatsink fins of most led lights are very ineffective because they are spaced too close together and not deep enough. Often times the fins are also next to the reflector. They need to be close the connection point of the LED to the housing though. This is only possible when the LED is not mounted too close to the part of the light that the users hand grips.
A 70W LED would be interesting in this host. It's designed more effectively than 99.9% of all large LED lights. For optimum use with an LED the reflector would need to be shortened and the head below the reflector would need to be solid aluminium or copper.
BTW: comparing the wheight if these lights only makes sense if you subtract the wheight of the battery packs. The battery cells and electronics can wheigh different amounts, but they don't dissipate any heat (only the metal housing).
I also spoke about the BLF-GT90 saying: “For the same reason the BLF-GT90 (2 Kg) maintains its turbo for 5 minutes whereas the turbo of the Acebeam K75, with about the same output, lasts just 90 secs.” ….
In order to better dissipate the heat the LED flashlights should have very large fins around the neck which unfortunately would be ridicolus.