Yeah I've spent a few hours going through all the logs (and writing stuff to make that possible on my little laptop screen), so without going into conspiracy theory mode, it's simply been a case of 2 orphans close together.

All our nodes accepted our block - i.e. none of our nodes around the world, that are all connected to the relay, saw slush's block before our block.
That is a significant point, since it means that the 2 competing blocks must have been very close together.
All the node block change times were 500ms or less, except for Silicon Valley was 2s, but that's one node in a group in that area, so all that matters is if one of the group processes it quickly and gets it to the relay quickly, which of course did happen.

Yet it would seem that all the well known sites saw slush's block before our block.

Bottom line is, there's that 500ms (or less) gap, which is almost certainly less than any of the larger pools that are confirming blocks properly, and during that time if another pool manages to submit a block to the relay, we 'can' be orphaned, which isn't of course impossible.