Added set_yaw_mode to better control of yaw controller changes and variable initialisation.
Replaced AUTO_YAW mode with separate yaw controllers YAW_LOOK_AT_NEXT_WP, YAW_LOOK_AT_LOCATION, YAW_LOOK_AT_HEADING.
Pilot manual override of yaw causes yaw to change to YAW_HOLD (i.e. manual yaw) until next waypoint is reached.
Added get_yaw_slew function to control how quickly autopilot turns copter
Changed YAW_LOOK_AHEAD to use GPS heading and moved to new get_look_ahead_yaw function in Attitude.pde
Renamed variables: target_bearing->wp_bearing, original_target_bearing->original_wp_bearing.
Removed auto_yaw_tracking and auto_yaw variables and update_auto_yaw function as they are no longer needed.
Simplified MAV_CMD_CONDITION_YAW handling (do_yaw). We lose ability to control direction of turn and ability to do long panorama shots but it now works between waypoints and save 20bytes.
Consolidated RTL state to be captured by rtl_state variable.
Combined update_RTL_Nav and verify_RTL functions which performed the same function but one was for missions, the other for the RTL flight mode.
Renamed some RTL parameters and global variables to have RTL at the front.
Landing detector now checks accel-throttle's I term and/or a very low throttle value
Changes include:
New low-level get_throttle_accel function takes target acceleration and compares vs earth-frame Z accelerometer values to produce output to motors.
Higher level throttle controllers modified to call new get_throttle_accel controller
Throttle_rate_stabilized controller added which maintains a desired climb/descent rate
Throttle_land controller added - descends using normal auto throttle controller to 10m then descends at 50cm/s
Multiple throttle modes added including landing mode
Land flight mode no longer needs GPS
Throttle cruise maintenance moved to update_throttle_cruise function
removed unused waypoint_radius var
removed slow_wp
updated speed management system to maintain a constant acceleration or deceleration of 1m/s
changed version to 2.8.1a
Also fixed a small bug in the get_stabilize_pitch function in which it was using the roll (instead of pitch) to decide whether it should let the i term build-up or not.
command_yaw_dir is an unsigned byte but was being set to a -ve number leading to very incorrect yaw target heading for counter clockwise turns.
command_yaw_end (i.e. turn's end target) was not being set if turn was of type "relative"
