This moves all RX packet processing into interrupt context, removing the queueing bottleneck that previous firmwares experienced. Before this change, we could handle at most ~3 packets before they would start NAK'ing due to lack of buffer space until the next main loop iteration pulled additional packets into the fbNew buffer. Now packets can immediately go into fbNew unless there's a pending frame finalization. This vastly improves the efficiency of the USB protocol, and we need far fewer NAKs.

I'd been assuming that the OS always copied our USB transfer buffer at submit-time. Turns out that this is true for Linux but not for Windows or Mac. This caused tearing when fcserver received a new frame while previous frame(s) were still queued in the USB host controller.

Fixed this by copying to a per-transfer buffer on Windows and Mac. Linux is unaffected. We avoid performing an additional copy there.

This was a doc typo, Fadecandy uses the standard OPC command number of 0x00.

Fixes issue 21.

Remove unnecessary columns, switch to .xls (it's what the assembly houses seem to want) and make the references consistent.

- Smaller apertures overall
- Larger apertures for LED
- Totally redo the LQFP apertures for U1

Changed the shape of pads for U1 and U3, trying to avoid having too much solder paste.

Missed a commit, oops.