Sure is a lot to go flamingly wrong there.
> there are really 3 power sections
I would say one power section, but made of three pairs. If you don't put anything in the output jacks, all the tubes are biased near-Off by the A B C returns to the 4K7 resistors at the output jacks. But there is only one transformer and its four secondaries are all paralleled together: this is a 1.33Ω amp with options to work at 2Ω or 4Ω at 2/3rd or 1/3rd of full power.
Most lesser tube amps are very un-fussy about load. 4, 8 or 16 ohms changes the way the bass resonance blatts, but the change in loudness is small and the tubes hardly care. This amp is worked so close to or over the limits that the load has to be RIGHT or things go wrong. Going a little higher Z is OK, but any lower or a lot higher will melt tubes or punch-through the transformer.
Note that G2 is fed 1/2 of full B+, typical for the huge power pentodes.
Note that, while oddly drawn, the B+ is a voltage doubler. This may dispel notions that doublers are always wimpy. Note that it has 315uFd of 700V filter cap....
The 6L6 is running 11 Watts idle, can deliver 4 or 5 watts of power to the grids. No, wait, it runs Triode, so only a couple Watts. While 6550 likes a low grid resistance and six of them gets very low, this does not make sense in Class AB1. They must be pushing significant grid current, grids swung up to +15 or +20V, grid current near 60mA peak, to justify the Watts of drive power being made. Driver transformer is possibly 2:1+1 ratio.
> I've never herd of anything like this before, it seems like a good way to destroy things.
AB2 and even full B2 drive used to be common when we strained to get a whole Watt, and also very common in kilowatt amps such as the 25KW amp that used to come with every big AM transmitter. But if you cheap-out on the drive, distortion can be obnoxious. And since the arrival of 6L6, it just has not been necessary to go Class 2 for big home or small gig work.
> And apparently you can't use regular 6550's or they will literally melt!
Actually, I would be leery of using any tube except the ones made FOR that amp. We are outside 6550/6550A ratings in so many ways, I would want tubes maufactured with the express condition that they would survive a Fender 400PS. A tube could meet or exceed all 6550A specs and still fail in this amp. I think only TungSol had the attitude and materials to make tubes for this amp.
OTOH, properly biased EL34 loaded to only 2.66 ohms (200 Watts) might live a long time. 1960s RCA 6L6GC might go as low as 2 ohms. It is the very low plate load impedance (and the unusual biasing) that will melt tubes.
> trying to squeeze 70 watts out of tubes intended to put out 35.
Properly run 6550 and 8417 will make 100 watts per pair reliably. It's in the book for 8417. The main problem is rail-sag, crossover distortion, and ignorant techs tying to bias-out the crossover distortion.
I used to run 450W-550W of tube amp, modded Bogens. Same 700V B+, and 350V on G2, but six pairs of 8417, not three pair 6550. Pure AB1, no trace of grid current (Bogen's lame driver would die before it delivered current). Which is cheaper: 6 extra ordinary-spec tubes, or that honking big driver, special protection schemes, and six special order tubes? Doubling the output tubes does mean more weight to heat the cathodes, but even with the hyper-hot 8417 I had no reliability trouble, and sure no melted tubes. You have to be careful around such amps, things go wrong in a hurry and very expensively, but this Fender is really pushing its luck.
Note that bias is specified as 35mA per tube, far below where most folks would set 6550 bias. Peak current per cathode is close to 500mA, and a 500/35= 14:1 range of current and over 3:1 variation in Gm ensures crossover distortion. If you do "the right thing" and bias for minimum crossover distortion, you will probably melt and implode tubes before you get there.
> he was in need of a bass amp
Then he is still in need of a bass amp. Without a current supply of the special-made 010309 tubes, and great understanding of its special needs, it is going to be a headache. I don't see a clean way to make it less insane. Re-labeling the output jacks as 8 ohms each would help, but who wants all this weight for just a lousy 217 watts? Sell it to the guys with the web page.