This works around an unexpected crash that would happen when you run the
following code:

```
cs = <EV3 color sensor instance>
# start sensor reset
cs.start_reset()
# wait for sensor reset and return new reading
cs.reflection()
```

The cs.reflection() call would raise SensorNotReadyError with the
message "Sensor has been unplugged". However, ideally, the call should
block up to 3 seconds until the sensor has rebooted.

It seems to me that the problem is caused by the fact that immediately
after cs.start_reset() is called, cs.is_ready() will still return true.
This confuses the waiting loop in read_raw_mode() - it will think that
the sensor has unexpectedly disappeared a ~millisecond later.

This commit fixes the issue by checking for the reset condition
in the is_ready() and similar functions.

I have missed that the sensor enters the "STATE_CALIBRATE_DONE" state
after the calibration. From the reverse-engineered sources it seems to me
that in this mode, the sensor will no longer process CMD-SELECT messages.

This commit should fix this by cutting off NACKs to the sensor and
letting it reboot.

It turns out that the underlying problem was that the sensor sends
the number of modes minus one and the code here did not account for it.

Major update of menu visuals and example programs

See merge request !12

This should resolve some licensing ambiguities - previously the code
had no license at all.

I am not entirely comfortable releasing the code under MIT (e.g. it can
then be used by a third-party vendor without giving any code back), but
MicroPython is under the same license and I want to keep the spirit
of that.