Another thought is that if you were using condenser mics, they could have picked up an airborne RFI/EMI spike. I dread to think it is the recorder, and the fact that it happens so infrequently makes it hard to reproduce and diagnose.
Were you using a memory card from the Fostex "approved" list? If not, that would be a clue. We know this click wasn't due to Windows or OS X trying to "do" something (e.g. wireless networking) and locking out CPU cycles.