Wow that thread is gigantic! I'm having a lot of trouble following the discussion. From what I can tell, liveforphysics agrees with me?
Let's use pronghorn's notation and speak in math terms. Sorry, it is my weakness here, but I think I have difficulty understanding the English. Equations though, I can understand them.
Adding one variable to the notation:
Q_max = maximum allowable rate of heat generation, which is a function of the max temp of winding insulators, the de-magnetization temperature of the magnets, the surface area of the stator over which to dissipate the heat, the color of the stator for radiative cooling, the desired service life of the bike etc etc. Practically speaking, this is an unknown value, but constant for a given make/model of motor regardless of how it is wound.
Starting from these relationships
1) Q_max = I_max^2 * R_m
2) P_max_mech = w_max * T_max
3) w_max = Kv * V_battery
4) T_max = Kt * I_max
5) R_m = constant / Kv^2 (assuming constant copper fill)
Now algebra:
I_max = sqrt(Q_max / R_m)
P_max_mech = Kv * V_battery * Kt * I_max
P_max_mech = V_battery * I_max
P_max_mech = V_battery * sqrt(Q_max / R_max)
P_max_mech = V_battery * sqrt(Q_max / (c / Kv^2 ))
P_max_mech = Kv * V_battery * sqrt(Q_max / c)
So in general, to maximize the system's power, pick the highest Kv you practically can. At some point you wont be able to fit the gearing to convert the power from high w, low T to a more useful low w, high T at the wheels, so you can't pick an ARBITRARILY high Kv. But as a general rule of thumb, you can err on the high side I think.