Your 32 pole, 36 slot machine has a winding factor of .945 when configured for 3 phase, when connected for 9 phases it will probably have the same. the theoretical lap winding of 3 slots per magnet pole has a winding factor of 1, in practice this is never achieved. but if your winding has a winding factor of say .86 (for a 24 pole 36 slot machine) it will only produce 86% of the voltage that a different coil configuration could theoretically produce with the same turns, and same volume of magnet. i have generally found this site to be pretty accurate
https://www.emetor.com/windings/the concentrated pole machines have significant third and higher harmonics and i suspect this will hold true for higher order phase connections as well... as the fundamental problem is that in order to get a sine wave machine you have lots of coils distributed around the magnet, or you have magnets tapered in ways to get a sine wave... but if you make a sine wave from distributed coils.. it adds copper losses. if you taper the magnet to produce a sinewave from a single coil, then you have removed magnetic flux from the machine that could have otherwise been flowing in the machine.
for example:
i rewound a very common 2 pole, 24 slot, 5KW "generator head" with a 3 phase lap winding with a pitch of 10. i took out 3 kilograms of copper and put 5 kilograms back in the machine. i was able to measure about 16 missing volts when comparing the line to neutral vs line to line voltage, representing third harmonic that was canceled out in the star configuration. you seem to be well aware of this with your measurements of shorting the neutral point to the line, for more resistance to turning forces. this is also why i realized a few years ago that the best test of a high torque low rpm generator is to short the windings and apply a torque and see how fast the shaft turns. if re-configuring the magnets for 9 phases produces slightly better results, then its better. technically, this has to be done after the rectifier.
so as for a 9 phase 32 pole machine, vs the same coils configured in 3 phase... which one is more efficient when driving a "dumb" 9 phase rectifier... that may depend on the specifics of just how much third harmonic is generated. it is true that the dc coming out of a 9 phase machine will be on the order of 3 times smoother than a three phase machine, but as far as the electrical efficiency of the generator.. it may not be improved at all.
much larger machines are designed to produce clean sine waves for a second reason: less eddy and hysteresis losses in the machine.
a small machine intended for simple 3 phase recification (or 9 phase) would likely do better to be designed for a trapezoidal voltage waveform.
most small BLDC motors for electric airplanes, scooters, ebikes.. are designed this way as far as i know. or it may be that its just simpler to drive them with a "dumb" 3 phase driver that simply pwm's the coil for a set number of amps, then turns the coil off and turns the next one on. making no attempt at all to push a sineusoidal waveform into the motor.
"dumb" rectifiers draw basically square wave amps from the machine, and a 3 phase trapezoidal emf generator will produce nearly constant dc current out of the rectifier, especially if an inductor is added after the rectifier.
a 9 phase machine required a third as much inductor (by energy, which means the inductor need be but half the size)