Just to clarify: Lost bytes have recently been identified and are now a known firmware problem.
It seems however that nobody has yet found the time to properly fix it. The problem seems to be extended periods of disabled interrupts in the USB code.
Several solutions have been proposed.
I always want to look at it, but then this is only one of way too many issues on my apparently ever-growing todo list. But then, it shouldn't even require me to get it solved anyway ;)