freeze_diagnose threshold: TAU_VIEW_STILL 4.0 -> 2.0 (more aggressive "is the scene really still?" check). Verdicts unchanged for current 41 events because they all trigger ot_loss via other signals.

figures/dataset_figures/freeze_diagnose/summary.json

@@ -5,7 +5,7 @@
     "ang_rad_s": 15.0,
     "opt_gap_s": 0.1,
     "dup_frac": 0.3,
-    "view_still": 4.0,
+    "view_still": 2.0,
     "tac_still": 3.0
   },
   "counts": {
@@ -876,5 +876,6 @@
       "opt_gap_s": 0.7798,
       "verdict": "ot_loss"
     }
-  ]
+  ],
+  "threshold_change_note": "TAU_VIEW_STILL lowered 4.0 -> 2.0 (anything above ~2 mean-abs-diff per pixel is visible motion; the prior 4.0 was lax). Current event set does NOT change because all 41 events also have other OT-loss signals (opt_gap / trans_vel / ang_vel) firing."
 }