In balanced operation the current flows from one amp through the headphones to the other amp, opposed to single ended where the current flows to ground, like this:

Single ended: [AMP (+) ] -> [headphone] -> [Ground connector]
Balanced: [AMP (+) ] -> [headphone] -> [AMP (-) ]

So with balanced there‘s no more current available then single ended, as the same current has to flow through both amps.
265 mA into 14 Ohm (28 Ohm balanced, see below) is ADI-2 Pro’s limit.

This is why in the lower diagram manual page 90, Current/Voltage vs. Impedance, there are no extra curves for balanced operation.
There is a difference, in balanced the impedance for maximum current would be doubled, so the limit is 265 mA into 28 Ohm.
Think of it as a doubled single sided connection with the headphones in the middle, with a “virtual zero, or ground” in the center of the headphones conductor.

Anyway, this is academic, as you cannot change your Susvara’s impedance.

The upper diagram page 90, Power vs. Impedance, provides all necessary information:
At 60 Ohm the red (balanced) trace shows 3000 mW.

How much louder would 6000 mW be, compared to 3000 mW?
Just 3 dB, not a night and day difference.

I found that ADI-2 Pro has sufficient power for even much “harder to drive” headphones, like my AKG K-1000: 120 Ohm, 74 dB / mW, and needs significant sub boost (which eats power) to sound good.

Guess what: I use it single ended on ADI-2 Pro and never run out of steam.

The power estimates you find all over the net are for real headbangers, or deaf people - which is the same anyway, as you soon get deaf with high sound levels.

In digital even headroom is of no concern, as there are no peaks above digital fullscale, and ADI-2’s meters clearly show how much headroom is left.
I always got plenty, even with highly dynamic classical music.

mA

current

Check the graph again. You have only 8 V at that impedance.

Thanks to you both for the replies, for some reason I thought the maximum output voltage was load independent as long as the current limit was not reached.
Even though it is quite obvious from the graph, I don't remember ever seeing this explicitly stated in the manual, I will go reread it

Close. The limiter is implemented via voltage comparator sensing the voltage drop of a thermal sensitive limiting resistor R, which does two things 1) current limiting: if the current passing through R is higher, it gets larger voltage drop. the comparator triggers.  2) overheat protection: R is also thermo sensitive. If the circuit gets warm, usually because you use it to drive a large load that consumes lots of power, R's resistance gets larger thus have higher voltage drop. the comparator circuit triggers.

You don't need to worry about the detail. The max power changes according to your room temperature, as discussed above, so it's not a fixed value.
All you need to know is ADI-2 has smart circuit design.
As long as it can drive your headphone sufficiently loud even in a warm room, you're good.

Hi KaiS, thank you for the additional information. It turns out that things are more complex than I thought and I'm learning more about the inner workings of amplifiers by the minute. It's not that I don't trust the charts in page 90, I'm just quite interested in why they are the way they are. 

This can be inferred from the charts but it wasn't completely clear to me that there is no simple flat line limiter that limits the current to 265 mA for any load, or limits the Wattage to 3W for any load.
There are only the limitations of the opamps which in balanced operation happen to max out at 265 mA for a 28 ohm load, and happen to max out at 3 Watts for a 64 ohm load and are otherwise perfectly described by the curves from the lower diagram of page 90.

Final question, is the limiting circuit in the OPA1688 chips any different from what would be found in a "normal" amplifier?

The classical power amplifier design has 4 stages of protection:

1. A simple overcurrent limiter that detects the voltage through the final stage‘s emitter resistors and clamps the drive voltage.
If not carefully designed, this one is notorious for causing power transistor failures when used with reactive loads like typical loudspeakers.

2. A fast blow fuse in the power output.

3. A thermal protection on the heatsink that either reduces or cuts the audio, or the PSU voltage.

4. A thermal switch inside the mains power transformer that cuts the primary.

The OPA1688, according to it‘s specs, uses advanced variants of 1. and 3.

RME added the circuit @ning described, and a 47 Ohms resistor in each of the 6 OPA1688 outputs that are used per channel.