The inductance of the power cable between the arc generator and the workpiece limits the rate at which the current can change.
The time it takes for the current to reach the desired level is: time=(desired current level / Voltage across inductor) * inductance of cable
The maximum output voltage of the BX17 arc generator is 150V. When arcing takes place, the combined voltage across the arc and ohmic losses in the wire contacts is roughly about 50V, which means that there is 100V across the inductance of the cable. The inductance of the power cable has been measured (as twisted pair with shorted end), and this is exactly 1 uH for a 2 meter length.
Now, the maximum current setting of the arc generator is 30A. This means that it takes (30/100)*1e-6=0.3 microseconds to reach 30A with a 2 meter long cable.
The shortest pulse on-time that the arc generator can make is 1 us. This means that during that 1 us, 0.3 microseconds is used as current rise time, 0.3 micro seconds is used as current fall time, and for 0.4 microseconds the current is actually at the desired level.
So you can clearly see, that the length of the cable affects the pulse rise and fall time. If the cable would be twice the maximum length, the inductance doubles, in that case the shortest BX17 pulse would be distorted too much.
If however, the BX17 is set for pulses that are longer (so machining at lower frequencies and/or higher duty cycles), than the distortion by the cable contributes far less and a longer cable can be used.
The main reason why a maximum cable length of 1.5m (1.5m twisted pair + 0.5 m untwisted for the C-arc = ~2m) is specified, is that only then it can be guaranteed that the pulse (on the workpiece side of the power cable) is within specification for the entire operating range of the BX17. Note that inside the BX17, there is also about 0.4m of cable length.