handle line segment termination in navigator

[tstastny]

Describe problem solved by this pull request
Waypoint item reached criteria for fixed-wing behaviors currently span the navigator, fixed-wing position controller, and L1 guidance library. Each location having slightly varying handling. For the case of waypoints in sequence, current handling strategies include:

• in navigator: check on horizontal acceptance radius + acceptance altitude (NAV_FW_ALT_RAD)
• in fw_pos_ctrl_l1: (for the case of loitering up to alt at the terminal waypoint) check on acceptance altitude (2 * _param_fw_clmbout_diff)
• in ECL_L1_Pos_Controller: once the aircraft is beyond the terminal waypoint and within some cone, guidance is switched to flying back to the terminal waypoint.
  

  PX4-Autopilot/src/lib/l1/ECL_L1_Pos_Controller.cpp

  Line 152 in 463513f

  } else if (fabsf(AB_to_BP_bearing) < math::radians(100.0f)) {

The latter condition was implemented to handle the corner case where the aircraft is "passed" the second waypoint, but not within the acceptance radius, and thus not caught for mission progression by the navigator.

Describe your solution
This PR attempts to begin some disentanglement of logic by

1. handling the "passed the waypoint" case in the navigator mission item reached check for fixed-wing vehicles following waypoints in sequence (line following) -- i.e. the line segment is considered complete when the aircraft is either passed the second waypoint OR within the acceptance radius (see figure below)
2. using the same acceptance altitude in both navigator and the loiter to alt logic in fw_pos_ctrl_l1 to avoid discrete clunky guidance behavior

NOTE: point (1) also offers the advantage that the specific L1 corner case handling is not necessary anymore -- which becomes relevant in #18810

Test data / coverage
The same logic was tested in SITL and can be seen #18810 (comment)

[sfuhrer]

he line segment is considered complete when the aircraft is either passed the second waypoint OR within the acceptance radius (see figure below)

This logic makes sense to me and certainly is clearer and more predictable than what is there currently.

Not directly related to the changes in this PR, just want to point out that the acceptance radius is coming from the L1 controller (_L1_distance), resp. Navigator takes the max(NAV_ACC_RAD, _L1_distance).

[Jaeyoung-Lim]

LGTM,

The only comment would be that L1 is also used by other vehicle types(such as rovers/boats) therefore we wont be able yo completely pull the logic out of the guidance logic with this approach