These are speculative thoughts:
I notice that the freeze frame data shows a single incidence of a misfire on #2 cylinder, which is in bank 2. I think I would dig into that further. Perhaps there is a slightly clogged fuel injector, or a marginal coil. If #2 is firing poorly, it could cause erratic lambda readings on that bank, and could cause occasional out of spec readings from the downstream O2 sensor.
This much I do know:
The downstream O2 sensor is there to ensure that the cat is doing its job. The cat can only do its job if it gets a stable stream of stoichiometricly burned gases as input. That can only happen if all 4 cylinders in a bank are operating properly and similarly, so that the upstream O2 sensor can tune the mixture properly. An out of spec or clogged fuel injector will throw a monkey wrench into the whole system, as will a marginal ignition pulse.
Another way of saying it is that the downstream O2 sensor is really the final check on the actual tailpipe emissions. As [MENTION=100496]blackworks[/MENTION] pointed out earlier, a downstream O2 sensor "failure" is often just saying that something upstream is out of whack; not enough to throw a code directly, but enough to cause an emissions failure at that instant. If you had a tail pipe sniffer measuring your emissions at the exact time of the code failure, my guess is the sniffer would have also shown emissions to be momentarily out of spec.