Interesting test results. You're showing higher voltage and current for Jerry-rigged than for Delta.
For a three-phase if the waveform is sinusoidal (or close to it, or any of a number of compatible waveforms) the voltage should be the same but delta should provide more current.
What this says to me is that the waveform is so far off sinusoidal that you have significant circulating current in delta and this is dropping the voltage (and burning power heating the mill, too).
Which is not necessarily anything wrong with the conversion. B-) It just means you'll want to do a few things:
1) Check that the coils are all intact: Does each phase, working alone, produce the same voltage and current as its brothers.
2) Check how you're connected for delta: Are you paralleling two sets of diodes in delta so diode drop doesn't unfairly distort the results? (Using bridges: Tie the two AC lines of each bridge to form one parallel set, for best balance.)
Presuming neither of those shows a problem:
3) Use Jerry-rigged rather than Delta if you want to do a high-speed mill or a delta-wye switch.
A nice thing about wye-jerry switching is you only need two contacts, shorting together the wye point without changing the rectifier configuration.
A downside to Jerry-rigged is that, if you want your rectifiers at ground level (which you DO if you want to do short-the-mill furling) you need six drop wires rather than three. (If you put the diodes up the tower you only need two, of course.)