Mike,
I took another look at the 915's schematic. I did notice one particular detail of the power supply implementation that we don't find often utilized in more recent products. Which is, that all amplification blocks appear to directly wired to the same +15V, -15V supply rails. I didn't readily find any provision even for passive decoupling of noise and inter-stage power supply disturbances, only local passive bypassing. It's pretty much standard practice today to find multiple independent supply regulators, or, at the least, passive decoupling being utilized to isolate supply rail induced disturbances between circuit blocks. I'm not sure there is much you do to change this without hacking up the PCB traces, possibly damaging the unit and rendering it worthless even to re-sell.
I can fully believe that the headphone section sounds none too pleasing. It is comprised of an single NE5532 op-amp per channel. It features a 150 ohm current limiting resistor, placed between the 5532 and the output jack. This effectively sets the output impedance at 150 ohms. This is too high for many audiophile headphones. Hafler might have utilized two or more NE5532s in parallel (many hands make for light work) to help deliver enough current to enable use of much lower value current limiting resistor, and increase total power dissipation capability. The additional cost and implementation trouble would have been minimal, I should think. On top of all that, the circuit appears to be connected directly to the same supply rails as every other gain block in the unit. Seems like incongruous cost based shortcutting, given the nice discrete JFET circuitry included elsewhere in the unit.