I am having a conversation with myself, but I hope this information helps someone else.
So interestingly I found that if I tie the UART RX or TX pins to ground, the current consumption drops to where I would like to see it, 1.3uA. Which seems to indicate that possibly an interrupt is being triggered on the floatiness of the UART pins. I can probably work with this by ensuring those pins do not float.
The odd thing is, that I do not have the box checked labeled "Enable wake from Deep Sleep Mode". So in my opinion if those pins or floating or not, nothing really should be happening as long as that box is not checked. Does that check box work on the PRoC 4 BLE chip?
Some further testing seems to indicate that any pin tied to an interrupt, will continue to be serviced to some extent in DeepSleep mode even if that interrupt is disabled. Others may already know this, but it is news to me. So please be mindful of floating pins that are tied to an interrupt or any digital component. If they will waste power.
Setting them to resistive pull down or resistive pull up (which ever is appropriate) appears to correct the problem.
A digital input left floating will potentially draw a lot of power.
This ap note discusses GPIO and other considerations when
sleep modes are used and issues associated with measuring
power while debugging.
http://www.cypress.com/?rID=78797 AN86233 - PSoC® 4 Low-Power Modes and Power Reduction Techniques
http://www.cypress.com/?rID=110007 AN92584 - Designing for Low Power and Estimating Battery Life for BLE Applications