An important difference between a CPU and an ESC is that a CPU actually does work. An ESC simply channels battery energy in a controlled fashion to the motor, which is the part that does the work. The heat in an ESC is only caused by FET switching losses and the rdson value of the FETs. Over time, this will improve as FET design improves.
Passive cooling has worked very well for me. I usually add
a little extra heatsinking in an area that will be exposed to at least minimal airflow. Adding larger heatsinks works, but you get diminishing returns if they get too big.
Solid state cooling is simply too inefficient, I agree. Condensation is not really a concern as long as you keep the temp at or above ambient. I don't think anyone has illusions of running an ESC as sub-0 temps.
And water cooling... um... yeah. I'm not gonna go there either.
Running ESCs in parallel would indeed help a lot, but the cost would be too high and the controller circuits would have to be designed to do so.