In-flight Barometer sanity checking will be a more complete solution but
until then this should catch some bad pre-flight barometer behaviour

Shortened gyro calibration commit also halved the total time we would
look for a good gyro calibration.  This restores the total time to 10
seconds.

when the airspeed ratio is far too low we were not raising it as the
airspeed was never getting above the minimum airspeed

Pair-Programmed-With: Jon Challinger

Trim save gives no feedback, so I've added a gcs_send_text() call to
print "Trim saved" so the user knows to re-center trims again

Removed delays before calibration, reduced number of samples taken,
widened convergence criteria

full rwtest on FMUv2, read test on FMUv1

using update_calibration() instead of the full calibrate() cuts
1.5seconds of the arming time

used to set PWM on motors when disarmed

this means we no longer store params on SD at all

This potentially changes the servo number that must be supplied if using
an APM1 but it now shares the same logic as arduplane and also adds
protection that the servo is not attached to the motor.
removed debug message from previous commit.

default behaviour remains as before, this is an compile time option only

defines.h is for defining enumerations and other unchanging values while
config.h is for changeable values like parameter defaults, enabling or
disabling sensors.

make it possible to build an APM2 specific build

be nice to release candidate users