Replace normal_triangle with normal_polygon for robust concave polygon handling

[Copilot]

Polygon.normal was calculating normals using normal_triangle with the centroid and first two points, which fails for concave polygons when those points lie on concave ears.

Changes

src/compas/geometry/polygon.py

• Replace normal_triangle([self.centroid] + self.points[:2]) with normal_polygon(self.points) in the plane property
• O(n) algorithm considers all vertices instead of O(1) triangle-based approach

tests/compas/geometry/test_polygon.py

• Add test_polygon_normal_concave validating L-shape and arrow-shape concave polygons with both winding orders

tests/compas/geometry/test_triangulation_earclip.py

• Update test_earclip_polygon_wrong_winding expected faces to reflect correct normal calculation for the complex concave polygon

Example

from compas.geometry import Polygon

# L-shaped concave polygon
points = [
    [0, 0, 0], [2, 0, 0], [2, 1, 0],
    [1, 1, 0], [1, 2, 0], [0, 2, 0]
]
polygon = Polygon(points)

# Now correctly computes normal considering all vertices
print(polygon.normal)  # Vector(x=-0.000, y=-0.000, z=1.000)
print(polygon.is_convex)  # False

The improved normal calculation also affects earclip triangulation orientation since it depends on polygon.normal to establish the transformation plane. The new triangulation is equally valid but differs in vertex ordering.

Original prompt

---

This section details on the original issue you should resolve

<issue_title>Polygon.normal with normal_polygon instead of normal_triangle?</issue_title>
<issue_description>Currently, the polygon's normal is calculated as the normal of one of its triangles: normal_triangle([polygon.centroid, polygon.points[0], polygon.points[1]]). However, this is problematic when the polygon is simple but concave, because the first two points might be on the concave ears, as pointed out by the comments inside polygon.plane.

The compas.geometry.normal_polygon seem to offer a more robust method of calculating such normal, with O(n) instead of O(1) for calculating the normal.

Is it possible to incorporate such method into polygon.normal, since the code is already available in compas? Or let the user choose if they would sacrifice computation efficiency for more accurate handling with concave polygons?</issue_description>

Comments on the Issue (you are @copilot in this section)

• Fixes Polygon.normal with normal_polygon instead of normal_triangle? #1487

---

✨ Let Copilot coding agent set things up for you — coding agent works faster and does higher quality work when set up for your repo.

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 62.16%. Comparing base (8221339) to head (9b320ce).
⚠️ Report is 13 commits behind head on main.

Additional details and impacted files

@@            Coverage Diff             @@
##             main    #1498      +/-   ##
==========================================
- Coverage   62.21%   62.16%   -0.05%     
==========================================
  Files         208      208              
  Lines       22437    22437              
==========================================
- Hits        13959    13949      -10     
- Misses       8478     8488      +10     

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.