Thank you for your recommendation of Steve Sakoman.  He was able to use his experience with the Omap and contacts at Gumstix and TI to perform a postmortem of my Overos and identify that my board was damaging the CPU, not the PMIC, indicating that the main power supply was not the cause.  Steve then zeroed in on my 1.8V power supply connected to the level shifters. This supply turned on at startup, but the enable pin should have been connected to SYSEN on the Overo to avoid power being applied to the IO pins before the CPU was ready.

Ultimately, given the number of expert engineers that looked over my design and missed this, my opinion is that the underlying problem was inadequate documentation.  Instead of stating that SYSEN must be used in this way, the documentation merely says "Connected to SYSEN on PMIC". Yes, the published schematics for the base boards have this pin connected, but there was no documentation indicating why or the reason this is required, and a design that only connected a few pins to the Overo would not even notice there was a problem.

In order to avoid future customers going through the same ordeal, I would respectfully suggest this matter be given some attention. At the very least a "lessons learned the hard way" section on the site detailing how the SYSEN, USBH_VBUS, and VBACKUP pins can cause damage in certain scenarios would be useful.  Detailing how OV_RX3 needs a resistor to prevent noise from interrupting the boot process is another example. 

Paul Nolan, CEO Idruna Software Inc.