Hm, I wonder exactly what they use. I would guess against gold doping -- Veb is pretty normal (7-12V), and storage time is massive (1-2us). Metal doped diodes of comparable rating (e.g., Pt FREDs) do a fraction of that (~100ns?) for the same voltage (say 600V). I'm... not sure how much difference it makes as far as BJT versus diode, for carrier lifetime.
Another factor, for or against gold doping, would be hFE at low currents, if it tails off quickly (recombination dominant) or stays pretty good down to low Ic. I don't have any data offhand.
I think it's more due simply to the modest doping of the emitter giving poor hFE, and the very low (and probably graded) doping of the collector / drift region, bordering on intrinsic, necessary to give the high voltage rating. A combination of reduced Rbb' (hopefully by interleaved base and emitter contacts, of whatever style), and probably something about doping profiles, allows reasonable rise/fall time (100-200ns) despite the relatively huge delay before getting there (i.e., storage).
On the upside, despite the low hFE, the power gain is excellent. A couple volts at the base, is able to switch a thousand at the collector.
At low power, design probably doesn't matter much, so long as hFE and fT are adequate. 2N3439 is an old one that, hm, good luck finding SOA or any other curves on it I guess, but I've used a few at voltage, in linear use (so, a few mA tops) without problems. Likewise with MOSFETs, there are plenty in small (SMT) packages that don't show 2nd breakdown if for no other reason than, they simply can't dissipate enough power to cause a fatal temperature gradient across the die.
Heh, on a less closely related note, I have found that HOTs can be avalanched in the same way as PN2369 and ZTX413 (and others) can, but they presumably do so in a localized manner, so are not able to carry any more current than the small types -- indeed, probably less, because they have to discharge their own junction capacitance through the avalanche site first. I've "blown a hole" in several, trying this. It's cool to see a transistor handle 1800V before making a nice sharp (sub-50ns) edge; unfortunately, they end up with a partial short of 20-60kohms C-E (after discharging ca. 220pF+), making them useless for further activity. (Other than the very high off-state leakage, they do function normally, suggesting a pointlike failure mechanism.)
Tim