My two cents:
High voltage is usually just a marketing issue. The gap ionization voltage (arcover voltage) is what determines the maximum voltage that the system sees under normal operation. If a plug wire falls off, then the coil will automatically keep raising the voltage (due to the collapsing magnetic field) until something, somewhere arcs over and dissipates the energy stored in the coil. The high numbers (50 kV) you see in some advertising are just the dielectric strength of the winding segments. It has nothing to do with the most important item, the energy in the coil.
A wide gap is theoretically desirable. A wider gap will light off a larger percentage of the air/fuel mixture, theoretically causing a faster burn (and therefore requiring less advance). Most anything that reduces the advance required will result in increased pumping efficiency (due to the piston not having to spend as many crank degrees compressing a burning mixture).
The downside of a large gap, in my opinion, is the increased sensitivity to having the spark blown out (or the "string" of ionized gas in the gap broken up easier) due to chamber swirl, requiring the coil to ramp up in voltage again to re-spark the mixture. This delay in resparking can cause a minor amount of spark retard, along with energy losses as the system has to recharge up a bunch of parasitic capacitances distributed throughout the system.
As there is no free lunch in an ignition system, whenever the gap is increased, the spark duration is decreased. The energy in the coil is consumed by gap voltage X gap current X spark duration. More gap voltage, less spark duration.
The gap current is affected significantly by the system resistance (the coil windings, plug wire resistance, and the gap impedance). I suppose high current will potentially put more instantaneous wattage into the gap, but the spark duration will decrease, as I mentioned earlier, and plug electrode life usually suffers.
As far as what level of energy is overkill, I believe it depends on each engine's "personality". I've worked with some other ignition engineers who were responsible for mapping out the ignition requirements of a new engine. They would vary the energy levels of their test setup and monitor the engine looking for performance dropoff, emission issues, etc, and then spec out a coil to be used on the engine, and the primary current, dwell times, and open circuit clamping voltages that the ignition module would have to control.
Again, just my two cents.