> make the cathode surface vary in shape under the grid to combat inselbildung.
Same problem which causes inselbildung in the first place. The ideal "grid" would be a plane surface. Grid wires can not be zero size, zero pitch, zero error, so can't be plane. Trying to correct their non-planeness with a lumpy cathode seems doomed. The grid errors are already at the fabrication tolerances, the complementary errors at the cathode are even smaller, and I don't see how if grid imperfections are a problem, cathode bumping won't be a larger (i.e. smaller) problem.
> if modern techniques
Well yeah. If you can write "IBM" on the point of a pin, as is done now by moving individual molecules around with an electron microscope, you could fabricate a much finer grid than WE could with highly refined versions of jeweler's techniques for working fine bits of metal. I believe modern CPU photolithography (electron beam?) techniques make finer structures than 416 grids. Flash aluminum on a wafer, mask a grid, etch the holes, dice to 416 size, then etch the silicon off the back. Walla, micron grid structure. Or cover the grid with thin SiO, etch most of it away, fill it with sugar, spray cathode oxide, flash with nickel, and wash the sugar out: complete grid-cathode structure with reproducible dimensions unimaginable in 1950.
> The phono pre probably was running at lower current...
Well, maybe just because guys who do 1mA-3mA on common receiving tubes can't wrap their mind around a 30mA tube. If you pencil 100V headroom above the plate, the plate resistor is 100V/30mA= 3K. Total power is like 10W. That's just strange for hi-fi work. And in fact there's no reason to drive a phono preamp down to 50 ohm noise resistance... it seems to be a moving-magnet preamp, pickup has ~5K dead resistance already. In part because the 5mV/47K pickup standard evolved for "good" noise performance with common Gm=1,000uMho (Rn=2K5) tubes.... not a lot to be gained with higher Gm.
> To install into the cavity properly required a special KS tool.... socket plug was non-standard too
Yeah, but for audio and DIY you would just use hose clamps and tack-solder the base pins. We aren't concerned with keeping the plate tank super close to the actual plate, or grounding the grid, or 99.9999% up-time among tens of thousands of units out in the field.