Webots sim-to-real fixes, DAgger pipeline, 360° proto variant

Today's session worked across the full Webots delivery stack — found and
fixed a cluster of bugs blocking the BC/RL transfer, then explored
training-side mitigations for the residual perception gap.

Bug fixes:
- Makefile FP_RATE default 2.0 → 0.0: BC demos used fp_rate=0 but RL
  fine-tune defaulted to fp_rate=2, poisoning the BC obs distribution
  and stalling PPO at 0% success across 1.46M+ steps.
- controllers/{shepherd_dog,sheep}/runtime.ini: Webots was launching
  controllers under system python3 (no numpy) and they were crashing
  silently. Pinned to the conda tir env.
- herding/config.py HERDING_WEBOTS preset: pen_latch_depth 0.5 → 2.0,
  max_new_tracks_per_step 3 → 1, static_reject 0.8 → 1.2. Stops phantom
  FPs near the gate from latching as permanently-penned tracks.
- herding/perception/sheep_tracker.py: penned tracks now decay at
  forget_steps × 8 instead of living forever. Adds get_positions
  min_freshness filter for deploy-time use.

Training/eval matches deployment:
- training/bc/collect.py: --dagger-policy flag for DAgger rollouts
  (policy drives, teacher labels) + --use-webots-preset for matched
  140° tracker + DR config.
- controllers/shepherd_dog/shepherd_dog.py: scan-fallback (0, 0.6) when
  BC/RL sees empty sheep_positions — recovers from FOV gaps.

Tooling:
- tools/dagger_round.sh: one-shot DAgger round (collect + concat + bc).
- tools/webots_sweep_gt.sh: full sweep with HERDING_USE_GT=1 for the
  perception-gap diagnosis matrix.
- protos/ShepherdDog360.proto: 360° FOV variant for the FOV-ablation
  comparison. Canonical proto stays at 140° per project spec.

Artifacts: v1 BC/RL policies for all 4 (drive × world) combos trained
in clean gym (success: diff/field 90-100%, diff/round 58%, mec/field
60-100%, mec/round 50-100%). DAgger r1/r2 BCs for diff/field show
12%→38% progression on gym HERDING_WEBOTS proxy but did not close
to actual Webots LiDAR (0/5 throughout). Next: LSTM policy or
learned tracker per the project-state memory.

Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>

tests/test_control.py

@@ -106,11 +106,34 @@ def test_sequential_empty_input_idle():
 
 
 def test_sequential_targets_closest_to_pen():
+    # With 2 sheep (≤ STRAGGLER_THRESHOLD), sequential goes straight to
+    # "targeted" phase and pushes the sheep nearest to the pen.
     near = (10.0, -5.0)       # closer to pen entry (11.5, -15)
     far = (-10.0, 10.0)
     sheep = {"near": near, "far": far}
+    vx, vy, mode = sequential_action((0.0, 0.0), sheep, PEN_ENTRY)
+    assert mode == "targeted"
+    # Dog should be directed toward near sheep (south-east), not far (north-west).
+    assert vx > 0 and vy < 0
+
+
+def test_sequential_collects_when_scattered():
+    # With >STRAGGLER_THRESHOLD sheep and radius > F_FACTOR*sqrt(n):
+    # should use collect (Strombom) not targeted.
+    sheep = {f"s{i}": pos for i, pos in enumerate([
+        (12.0, 10.0), (-12.0, 10.0), (0.0, 12.0),
+        (12.0, -12.0), (-10.0, -8.0),
+    ])}
     _vx, _vy, mode = sequential_action((0.0, 0.0), sheep, PEN_ENTRY)
-    assert mode.startswith("drive:near")
+    assert mode in ("collect", "drive")
+
+
+def test_sequential_drives_when_compact():
+    # Compact flock of 5 sheep near centre — should drive, not collect.
+    sheep = {f"s{i}": (float(i) * 0.3, float(i) * 0.3)
+             for i in range(5)}
+    _vx, _vy, mode = sequential_action((0.0, 5.0), sheep, PEN_ENTRY)
+    assert mode == "drive"
 
 
 # ---------------------------------------------------------------------------