This latest evaluation is really great news!
I'm still confused about the continued need for thermal calorimetry though. What is preventing the use of rectification and filtering of the output pulses into a DC output? Advanced silicon rectifiers have reverse recovery times of <40nS at many kilovolts with just a few picofarads of junction capacitance.
Does an excess of reactive impedance in the load quench the reactor output? If so, has a threshold of this sensitivity to reactance been characterized? Surely there is circuitry that can be applied to harvest this energy without thermalization. Since the output waveform seems to follow a well-characterizable form, perhaps a dynamic impedance response could be applied to present an optimized load impedance to the reactor output stage if necessary. The length of single-ended wiring in the output stage of earlier demonstrations suggests a tolerance for nontrivial levels of inductance at least.