IIRC NEMP is assumed to be at 50kV/m for milspec certification purposes. However most exposed devices are much simpler than quadcopters. There are some class exemptions too, e.g. electronic sights are allowed a fraction of second blackout after upset. Not something that would work for an attack drone though.
What I think makes Leonidas more efficient is they likely operate in continuous wave bursts rather than pulses. Probably with a broad comb rather than one specific value too.
I don't think a fraction of a second interruption would be overly problematic. These drones are somewhat heavy and thus bring relevant amounts of momentum. But the start up process is probably far too long
Yes the reboot takes too long, even before you account the time to re-negotiate comms (even over fiber it is not instant). However in a terminal approach on target say a 1/4 second disrupt will likely be unrecoverable. Quadcopter drones are not ballistic or aerodynamic, and stability recovery once it gets tumbling is challenging.
The longer pulses the in this platform seem to be a big part of delivering effect on target. Area under the curve is where the damage happens.