I'm sorry I don't have anything to add except how awful I feel about all of this. One thing that is clear, though - our statistical models are chasing climate change and unfortunately appear to not be equipped to catch up. We're using what is called Atlas 14 (NOAA's newest prediction model for precipitation) and it falls way short of the rainfall totals we're seeing in localized areas.
Is it possible that these events are way outside the normal probabilities of exceedance? Yes, it's possible. It's not statistically likely given the frequency of exceedance, though.
As I was telling a licensed engineer who was revising a (500 year!) floodplain the other day, you're just weighing the odds of a larger event against the cost of increasing fill/conveyance. It's ultimately not going to matter if the design storm is inadequate; the fault will lie entirely on whoever is promulgating a maximum exceedance event as less than its actual volume.
Engineers can only do so much with the data provided, and climate change is making it so the backwards looking statistical models are just not adequate in a lot of cases.