I've been playing with the waveform file and getting pretty much nowhere. I saved a waveform that says it has 1400 pts (50ns/div, 2GSa/s). The file is 22,376 bytes long and has a bunch of stuff that makes no sense to me. But if I read the data in as 8-bit bytes I find the trace stuck on the very end as two interlaced 728 point traces (yes it's 728, not 700). By interlaced I mean one trace followed by 4 bytes then a second trace. When I stick the two traces together (i.e. deinterlace them) it almost exactly matches the screen capture of the waveform. I have no clue what this means, but the data I was able to find was straight 8-bit bytes. (That doesn't mean that other versions of the data aren't hiding elsewhere in the long file.)
Also, if you save data taken in Hires mode to CSV, the results in the CSV file are quantized at 8-bit (as I mentioned in my original post). (The Hires mode does give enhanced resolution on the scope screen readout.)
One bug I found (or re-found actually) on my scope (FW 00.00.01.00.02): if I try to save waveforms to a USB stick without first saving a screen capture using the print button, my scope hangs and must be powered down. If I save a screen capture first, after each powerup, then saving waveforms and CSV work fine. Tested two USB sticks.