some thoughts and calculations lead me to believe that
(a) the exit pulse is probably on the order of 10 ns;
(b) the inductance of the coil should be about 30 uH overall;
(c) the rogowski coil should use superconducting YBCO tape in a liquid N2 bath. each 1cm wide YBCO wire can support 300 A current; (though i have not yet made allowance for the magnetic field)
(d) transducer diameter is 3 m, coil inner diameter is 0.955 m, outer diameter is 6.955 m;

this alters the design to 2.22 MV and 75 kA for 10 ns, or 250 coils of 5 turns each, in parallel. (total 1250 turns).

i expect efficiency to be above 88%, so long as the coils are tightly coupled magnetically (> 0.999), and the parallel capacitance is below 20 pF.

the following image shows what the coil will look like with about 1/3 of its windings on: http://qikr.co/files/pics/r/rogowski4065981.png