Mecanum Webots via Supervisor kinematic injection

Replace the failing ODE-rolled mecanum chassis dynamics with a
Supervisor.setVelocity call that uses the gym mecanum forward
kinematics formula directly. Wheel motors still spin (visual);
chassis motion comes from the gym model so training and deployment
match by construction.

Results (seed=42, n=10 sheep): BC + RL mecanum pen 10/10 in both
field and field_round. n=5 mecanum cells still 0/5 due to tracker
phantoms anchored to wall corners under the 360° LiDAR — documented
in docs/status.md as the remaining gap.

Cleanup: drop deploy-time hacks (HERDING_HEADING_*, HERDING_OMEGA_CLAMP,
HERDING_TRACKER_*) that were workarounds for the old ODE chaos;
revert the proto inertiaMatrix, roller dampingConstant, and reduced
motor torque since they no longer carry load; refresh comments
around the mecanum config presets.

herding/config.py

@@ -266,18 +266,21 @@ class RobotConfig:
     strafe_efficiency: float = 1.0
     """Mecanum strafe magnitude as a fraction of textbook X-pattern.
 
-    ``1.0`` (default) = perfect mecanum kinematics. ``0.4`` matches the
-    Webots roller-hinge mecanum proto calibration (62% slip on strafe,
-    11% on forward). Used by ``mecanum_step`` only — has no
-    effect on differential drive.
+    ``1.0`` (default) is the ideal kinematic mecanum. Values below 1
+    model strafe slip; the Webots controller reads the same value and
+    applies it in the Supervisor velocity injection, so gym training
+    and Webots deployment see identical body motion. No effect on
+    differential drive.
     """
 
     strafe_to_forward_bleed: float = 0.0
     """Fraction of ideal strafe magnitude that bleeds into body-frame x.
 
-    ``0.0`` (default) = no bleed. ``-0.28`` matches the Webots proto's
-    consistent backward push under strafe commands. Used by
-    ``mecanum_step`` only.
+    ``0.0`` (default) = no bleed. Non-zero values add
+    ``strafe_to_forward_bleed * |vy_body_ideal|`` to ``vx_body`` to
+    model the consistent forward (or backward) drift that some
+    mecanum chassis exhibit during pure-strafe commands. No effect on
+    differential drive.
     """
 
     def __post_init__(self) -> None:
@@ -407,23 +410,49 @@ HERDING_MEC_WEBOTS = HerdingConfig(
     ),
     robot=RobotConfig(
         action_smooth=0.55,
-        strafe_efficiency=0.4,
-        strafe_to_forward_bleed=-0.28,
+        strafe_efficiency=0.26,
+        strafe_to_forward_bleed=-0.40,
     ),
 )
-"""Webots-mecanum-matched training preset.
+"""Mecanum + 140° LiDAR preset.
 
-Same as HERDING_WEBOTS but with the gym mecanum kinematics scaled to
-match the Webots roller-hinge mecanum proto:
-* ``strafe_efficiency=0.4`` — strafing produces ~40% of textbook
-  X-pattern lateral velocity in Webots; this matches the bias.
-* ``strafe_to_forward_bleed=-0.28`` — strafe commands bleed ~28% of
-  their magnitude into backward body motion in Webots.
+Mirrors HERDING_WEBOTS but with mecanum-specific kinematic scaling
+(``strafe_efficiency`` and ``strafe_to_forward_bleed``) applied to
+the gym forward-kinematics formula. The Webots controller reads
+these same values via ``RobotConfig`` and feeds them through the
+Supervisor velocity injection, so gym and Webots produce identical
+body motion. Diff-drive ignores both fields.
+"""
 
-Use this preset when training BC/RL for the mecanum drive so the
-policy learns to compensate for the imperfect physical mecanum.
-Differential drive ignores both parameters and behaves identically
-to HERDING_WEBOTS.
+HERDING_MEC_WEBOTS_360 = HerdingConfig(
+    lidar=LIDAR_WEBOTS_360,
+    # Looser detection thresholds for the wider FOV — the 360° scan
+    # catches far walls, gate posts and pen rails the 140° front cone
+    # never sees, so the cluster/feature filters need slightly more
+    # margin to keep promotion rates similar.
+    detection=DetectionConfig(wall_reject=0.6, static_reject=1.2),
+    tracker=TrackerConfig(
+        forget_steps=300,
+        max_new_tracks_per_step=2,    # 360° gives more candidates per step
+        pen_latch_depth=3.0,
+        predict_steps=180,
+        consensus_k=3,
+        consensus_radius_m=0.3,
+        consensus_max_age=20,
+    ),
+    robot=RobotConfig(
+        action_smooth=0.55,
+        strafe_efficiency=0.26,
+        strafe_to_forward_bleed=-0.40,
+    ),
+)
+"""Mecanum + 360° LiDAR preset (the deployable mecanum target).
+
+The 360° FOV gives the policy perception coverage in every direction,
+which matches the omnidirectional motion the mecanum chassis can
+produce. Used for both gym training and Webots deployment so the
+trained policy sees the same observation geometry it will face at
+deploy time.
 """
 """Webots-matched training preset.