I think they do disagree. GPIO in this context is a specific type of I/O, as it is listed separately from PWREN# and SLEEP# in the same table. Hence, PWREN# is not GPIO.
I doubt that they would have been reconfigured and then returned to the supplier, though not impossible, I guess. I have two different ones that were in different factory sealed packages. Both have the same configuration or lack thereof.
Anyway, I'm more confident that the configuration that was read in from the chip is not corrupted, because I saved the config file to disk and then read it in an xml editor. The parameters all look reasonable. So, I'll probably take a chance and try reconfiguring the chip.