The high voltage simply overcomes ESR better. If you reduce ESR, then there's no need for a high voltage, unless the flash-tube needs to be made lengthier for some reason. Most flash units these days use no more than 350 volts on the storage capacitors. Otherwise controlling the flash pulse with IGBTs or Thyristors gets expensive. The ridiculously costly Profoto Pro-10 offers 2400 Joules with a t0.1 duration of only 2.5 ms at full output. So flash duration definitely isn't proportional to output energy. It's only when the flash exposure is regulated down from full 'power' by using electronic control that the duration gets shorter. Exactly in the same way that the camera shutter speed works. Except the flash duration has to be regulated to work according to the flash tube's exponential rise and fall curve. And that's why I showed the integral exposure curve on the graphs above.