feat(geofence): add segment geometry helpers

[JonasPerolini]

The mission checker validates individual waypoints but misses geofence breaches between them. This stack of 5 PRs adds checks at mission upload to catch those breaches before flight. To test the complete feature, use the last PR in this stack: gf-jump.

Algorithm:

1. Check that the first waypoint is inside of all GFs (validated)
2. Check if the line between waypoints crosses any GF, if it does, we are now outside of the validated zone --> breach. Boundary contact counts as a breach to ensure that a path that exists the GF through a vertice is caught.

Note: The implementation is differemt to the GF-aware RTL planner. The RTL planner’s checks are designed to find routes around fences. They use prepared polygons in local coordinates and allow paths to touch corners or follow their edges. Mission validation uses a simpler rule: once the starting point is valid, touching any fence boundary is a breach. This lets us check polygons directly from the existing fence cache. It avoids adding another projected fence (in local coordinates) or changing the RTL planner’s boundary rules.

This PR adds segment intersection and point-to-segment distance helpers for mission geofence checks. No behavior change, only lib.

[dakejahl]

Claude review on behalf of @dakejahl

nit, src/lib/geofence/GeofenceUtilsTest.cpp:96: the intersection table has no non-degenerate, non-collinear pair where a side value is exactly zero and the answer is still false, i.e. an endpoint on the other segment's supporting line but beyond it, with overlapping bounds: {"EndpointOnLineBeyondSegment", {0, 0}, {4, 0}, {5, 0}, {3, 2}, false}. That is the configuration where the inclusive <=/>= in segmentsIntersectInclusive (geofence_utils.cpp:535) accepts the first straddle test and only the second one rejects; the zero-side rows in the table today are all touching, collinear or point cases. One more row pins it.

[JonasPerolini]

Thanks for the review. Test case added here e63e3be @dakejahl

[github-actions]

🔎 FLASH Analysis

px4_fmu-v5x [Total VM Diff: 0 byte (0 %)]

    FILE SIZE        VM SIZE    
 --------------  -------------- 
  +0.0%     +84  [ = ]       0    .debug_abbrev
  +0.0%     +40  [ = ]       0    .debug_aranges
  +0.1%    +272  [ = ]       0    .debug_frame
  +0.0% +7.81Ki  [ = ]       0    .debug_info
  +0.0% +1.55Ki  [ = ]       0    .debug_line
    [NEW]      +7  [ = ]       0    [Unmapped]
    +0.0% +1.55Ki  [ = ]       0    [section .debug_line]
  +0.1% +2.52Ki  [ = ]       0    .debug_loclists
    [NEW]      +3  [ = ]       0    [Unmapped]
    +0.1% +2.52Ki  [ = ]       0    [section .debug_loclists]
  +0.1%    +358  [ = ]       0    .debug_rnglists
  +0.0%    +414  [ = ]       0    .debug_str
    [NEW]      +1  [ = ]       0    [Unmapped]
    +0.0%    +413  [ = ]       0    [section .debug_str]
  +0.0% +13.0Ki  [ = ]       0    TOTAL

px4_fmu-v6x [Total VM Diff: 8 byte (0 %)]

    FILE SIZE        VM SIZE    
 --------------  -------------- 
  +0.0%      +8  +0.0%      +8    .text
    +0.0%     +12  +0.0%     +12    [section .text]
   -30.8%      -4 -30.8%      -4    g_nullstring
  +0.0%     +84  [ = ]       0    .debug_abbrev
  +0.0%     +40  [ = ]       0    .debug_aranges
  +0.1%    +272  [ = ]       0    .debug_frame
  +0.0% +7.81Ki  [ = ]       0    .debug_info
  +0.0% +1.54Ki  [ = ]       0    .debug_line
   -20.0%      -1  [ = ]       0    [Unmapped]
    +0.0% +1.55Ki  [ = ]       0    [section .debug_line]
  +0.1% +2.53Ki  [ = ]       0    .debug_loclists
  +0.1%    +358  [ = ]       0    .debug_rnglists
  +0.0%    +450  [ = ]       0    .debug_str
  -0.1%      -8  [ = ]       0    [Unmapped]
  +0.0% +13.1Ki  +0.0%      +8    TOTAL

Updated: 2026-09-22T11:05:16