In that case, you're absolutely correct, it's operating on the edge. The Philips data sheet doesn't show that limitation, and maybe the AR designers didn't see it either. In any event, the intention was obviously to squeeze as much drive capability out of the 6DJ8 as possible. At this level though, they should have used a different tube.
Here's the simplified schematic with all the extraneous circuitry removed:
I modeled this in SPICE and made a few changes to the cathode followers. Simply increase the value of R13 to 180K, and increase the value of R14 to 39K. Both new parts should have a 1W rating. This will reduce the dissipation in the followers to about 70% of the original value and extend their lifespan considerably. The reduction in current of course means less drive capability, but it will still drive 18V RMS cleanly into a 10K load.
This change will also reduce standing DCV on the cathodes to about +100V. The heaters are elevated to about +122V, so this is well within the H-K ratings for the tubes.
