Fix many quirks in USB implementation

[elfmimi]

Prior to the fix:

• USB enumeration was quite unreliable especially on Windows.
• usb_serial_echo.rs example was occasionally dropping characters.
• usbreset command on Linux was unresponsive and in most cases BUS/DEV number differs after reset.

[michalfita]

Could you please comment changes better. The original code comes from Grepit and had been battle tested to certain extend, but as we're getting extra hands on these, this aspect needs improvement.

@martinmortsell would you have chance to test this in any way?

[tmplt]

Glad to see this module getting some much-needed love!

From a brief look I recommend more documentation (in general), but perhaps a layer of abstraction above the register modifications. My knowledge of USB and the peripheral itself has bit-rotted, so I cannot help in review efforts, unfortunately.

[michalfita]

I have 2 ATSAMV71-XULT evaluation boards, but they have pre-sales chips on them. New SAMV71Q21-AAB cost £18 each w/o transport and resoldering. I could test this but I need some funds for making the hardware up and running.

[martinmortsell]

Sorry for not responding, I have a V71-XULT myself, I'll try to make time for testing this during the week.