this fixes several problems with reading analog sources:
- we were getting poor values because we didn't wait long enough for
an analog source to settle
- we wasted a lot of CPU cycles waiting for conversions
- we were not taking averages over many samples, which we did with
the old AP_ADC driver on the APM1
this library abstracts out the way of getting an analog value. If the
ADC library is being used then it calls the ADC Ch() method, otherwise
it calls analogRead()
Andrew Tridgell
Randy Mackay
Pat Hickey