I did the reverse phase test again but got same results, however -
I ran the regular thd and imd tests with lower signal levels and lighter loads and it did make a big difference to the IMD, (but not as much to the THD).
The weird IMD disappeard when I lowered the volume of the amp as well as when I reduced input signal going into the amp. For this test I even reduced the load to a purely resistive 200ohm load (resistor). But do note that these volume levels and loads are ridiculously light load for an 'amp'.
Also, the volume level at which there is no imd with the resistor, the same imd exists when I simply change it to the headphone - indicating that the reactive and higher load is causing the imd.
It still doesn't prove for sure that its the opamp's drive capability, it could be the supply rail capacitance or its impedance, because of the fact that the Headsix uses the single version of the same dual opamp and does not suffer from this issue.
On another note I measured the input impedances of both the amps. The GV6 came out at ~9k which is pretty standard, but the Headsix unlike what it says on its website wasn't even close to 10k, it came out somewhere around 66k! Not sure what that is about?!