Ok, so I've done a lot of thinking about this and I'm pretty sure I got this figured out. Given two different setups:
1) 150 kv motor, geared 24:36 with 83mm wheels, 6s LiFePo, theoretical top speed of 19.2 mph
2) 300 kv motor, geared 12:36 with 83mm wheels, 6s LiFePo, theoreitcal top speed of 19.2 mph
Since motor torque constant = 1/Kv, if you do the math you find that:
1) motor torque constant = 0.0637 N*m/A
2) motor torque constant = 0.0318 N*m/A (half of motor 1)
However, your gearing ratio makes up for that, so torque at the wheels is just
1) torque at wheel = 0.0955 N*m/A
2) torque at wheel = 0.0955 N*m/A (the same)
Now, electrical losses are proportional to the electrical power put into the motor, which is given by P = I*I*R.
The way the manufacturers actually achieve the different kv ratings is by changing the number of turns in the motor. Motor 1 has twice the number of turns as motor 2. Obviously, this requires wires that are twice as long, but to fit the extra length they have to use wire that is exactly half the cross sectional area. Since the resistance of a wire is proportional to Length/Area, as long as the wires are made of the same material, the two motors have the exact same electrical resistance.
And since both motors draw the same current per unit torque at the wheels, they consume the same electrical power to move the same mechanical load and therefore identical electrical losses as well.
What does all this mean? To my eye, I would say that for a given line of motors (say the Turnigy SK3), and assuming you gear it properly:
1) Kv has no impact on wheel torque and speed assuming you gear it properly
2) Kv has no impact on electrical efficiency
3) Pick a Kv that stays under 8k rpm at max speed to avoid electrical inefficiencies not covered in this summary
4) Pick a lower Kv whenever possible since your mechanical drivetrain will likely be more efficient at lower speeds
5) Pick a motor whose Kv value works well with the commonly available pulleys/belts to get you the top speed you want.
6) If you're running an unsensored setup, dont pick a motor with a Kv that is too low or you'll have more issues with cogging when launching from a standstill (since our ESCs have trouble sensing the position of a VERY slowly rotating motor). Frankly though, I'm sure this is basically completely negligable.
This doesnt tell you anything about how to compare different lines of motors (i.e. SK3 vs NTM vs EMP motors) because across brands there may be other design (stator diameter, shaft supports etc) or quality differences (bearing quality, copper alloy) that will affect things. To make that decision, we'd really need to empirically measure the motor torque constant since and look for the motor that produces the highest torque per unit amperage. But at least if you've decided on the brand of motor you want to use, this should help you decide what version to buy.