Sat Sep 7 11:32:51 EDT 2013

USB fix: >63 byte descriptors

It looks like this is so hard because I don't really have a good way
to test it: the system is talking to the hardware in a way that cannot
be separated/simulated currently.

Maybe the way to go is to split that first?

I feel there is some ground to gain here, but I'm lacking insight..

Next: what can be tested is an interface that sends data incrementally
from flash.  I.e. the state machines themselves.