In the case where electrolytics are present, it will remove the leaked goo. The leaked goo is a double whammy. It is corrosive and can eventually destroy parts of the logic board. The goo is also conductive, and can cause short circuits on the board. This is why washing the boards will sometimes get it working again -- for a while.
In your case, with tantalums, even if the caps have died of old age, there should not be leaked goo on the board, so a wash should not, in theory have any effect. It might be an interesting experiment to try any way. Always good to test our hypotheses with experimentation.
As far as your bizarre hard drive/boot behaviour, beats me. Maybe try a formatter other than the hacked Apple formatter. I was always fond of APS's Alliance Power Tools. Also double check that your SCSI chains are configured properly with no ID conflicts, nor termination errors.
BTW, you don't have any of the internal drives set to the same ID as the external MO drive, do you? If you do, that's likely to be your problem.
The Q950 does not have SCSI Manager 4.3 in ROM, so it treats the two SCSI busses as a single logical bus, and SCSI ID conflicts will happen at boot time if devices share the same ID.