Scaling of bottom half of throttle was changed from THR_MIN ~ 500.
Previously it was from 0 ~ 500.

Accel throttle's I term is taken from scaled manual throttle

Release Notes updated

A new radio frame could arrive just as the frame time vs current time
check was being executed causing a false positive

Also bug fix to hold yaw on take-off

set_next_WP function was not setting the alt_change_flag properly

This indirectly switches surface tracking on/off which is useful to
avoid tracking the tops of uneven objects when far above the ground

Copter: constrain loiter so that it never commands more than 45 degrees of lean in lat or lon directions

APM_Config.h should be used to overriding the standard parameters but
they still require a definition in config.h or they will be undefined
when built by the mission planner.

Previously we checked if the alt > target_alt but because there is
little to no overshoot with the new accel based alt controller this
check was failing for some users.

An hidden bit of failsafe functionality in the set_mode function was
switching the throttle mode back to THROTTLE_AUTO instead of
THROTTLE_LAND

new set_accel_throttle_I_from_pilot_throttle function copies the
difference between the pilot's throttle and hover throttle to the
accelerometer based throttle controller's I term.

This effectively means that an RTL kicked off by a failsafe will
immediately switch to LAND mode only if within 2 meters.  Previously the
radius was much wider (15m).

This message takes 1ms to write and we are running over our limits on
some loops.

Jonathan Challinger added

this ensures the rate controllers use the latest data

this avoids problems in the scheduler turning up as errors in the DCM
calculation. The MPU6k produces samples with more reliable timing than
our scheduler does!