The desired_accel is set again 11 lines lower so this line did nothing.

Scales filter frequency using EKF nav gain scaler to take advantage of the reduced nav frequency whichcan tolerate a slower angle response.
This is required to reduce the effect of EKF optical flow noise increase with height.

Allows gains to be adjusted to compensate for optical flow noise

This has been done to provide a smooth psotion hold when using an optical flow aided EKF which can be noisy.

Signed-off-by: priseborough <p_riseborough@live.com.au>

fixes issue in which now could be earlier than _last_update_xy_ms
leading to a large dt value and a sudden lean on takeoff

Replaced hard-coded filter with LowPassFilter class allowing the
filter's to be 2hz on both APM and Pixhawk

This avoids an instantaneous jump in throttle during take-off by loading
the accel PID's I term with the expected change in throttle level

Velocity controller interpretsthe velocity requests as
desired velocities (i.e. feed forward).  These are then used to update
the target position and also added to the target velocity.
Also renamed the set_desired_velocity() function to
set_desired_velocity_xy() to make clear only lat and lon axis are
updated.

lean_angles_to_accel is now used to load the position rate I terms based
on vehicle's initial roll and pitch angle

Also removed _sin_yaw, _cos_pitch, _step saving a total of 17bytes of
RAM
Also made get_speed_up, get_speed_down, lean_angles_to_accel functions
const

init_xy_controller also added to capture initialisation all in one place

Pair programmed with Randy

Consolidated z-axis timeout checks to save 4bytes of RAM
Added POS_CONTROL_ACTIVE_TIMEOUT_MS to make timeout consistent

Fixed bug in get_stopping_point_xy in which it would update Z-axis
target if vehicle was moving less than 10cm/s horizontally

Stops horizontal PID's I terms from being reset when the controller is
next updated