Jamie,
You seem to have 3 threads going on about this motor. I am avoiding the other one because it seems to have turned into the kind of shitshow one might expect on Jeremy Kyle... Could I recommend you limit your thread proliferation and umm... treat the internet like the fragile ecosystem of souls it is? Of which we are merely two extremely fragile egos among a billion others.
I am still interested in this, since it is an actual attempt at doing something among the million "why doesn't my motor spin" threads. It is also nearly the motor I have (I have the external controller one).
How did the motor rewind end up? how much copper did you pack in?
Re. the controller, current etc, I expect your controller is running BLDC, square wave, 6 step, 2 phase at a time mode.
You get better performance from a sinusoidal/FOC controller, which runs all 3 phases at the same time. The mag field from the stator is then a continuously rotating vector, rather than one with 6 discrete steps per eletrical rev. With losses proportional to I^2, running 3 phases rather than 2 with the same overall power can be much more efficient.
It will also be quieter. Sadly I cannot recommend a good controller, they are all fraught with issues - ASI is virtually unobtainable, Nucular is huge lead time, VESC based is kind of rubbish; the project leader seems to have minimal interest in ebikes, and has been sitting on much needed fixes and improvements to the motor control code for years, my controller is an educational toy for me, the Alien ones are meh... Sabvoton seems to get a lot of users with minimal fuss and seems less prone to exploding than Kelly.
Regarding the battery vs phase amps, the motor and controller acts as a crude buck-boost converter, like a DC transformer... which converts [essentially...] battery volts and amps into phase amps and volts while conserving power (-losses to resistance and transient errors to inductance). Where a transformer does this by the turns ratio, a buck boost does this by varying the duty cycle and using the motor coil inductance to store and release the energy.
Vbat*Ibat= sum(Vbemf*Iphase)+sum(Iphase^2*Rphase)+sum(I*Ldi/dt) + Iphase*duty*Rbattery
Where sum(...) denotes the sum across 3 phases. For a 6 step this is easy, just use the ones of interest, for a FOC or sinusoidal it is harder and is done via a series of matrix transformations into 2 phase.
There is ultimately a hard limit on the motors abilities given by the copper fill (and iron saturation and magnet strength...), and rewinding does not show up magic combos, only incremental improvements from slightly better fill, less space loss to insulation, thinner wire for less skin effects.... and yes, you could possibly run it hotter with higher grade wire, provided you don't start to damage the magnets or other structural components.
Consider that (simplified, not accounting for reluctance torque):
T=[constant for geometry]*BstatorXBrotor (X denotes cross product)
Brotor is dependent on the magnets.
Bstator is given by the amp turns. More turns gives more torque, BUT the turns area decreases, and they get longer, so more resistance. The Back EMF of the motor also scales with number of turns, with
V=[constant for geometry...]*n*d(phi)/dt.
Overall, there is no net gain, only potentially better matching to your controller's capabilities, or better scaling of the speed torque tradeoff to your use case.