revert(ml): keep head auto-apply precision at 0.97 (operator: general tuning was fine)

Milestone 139 raised head_auto_apply_precision 0.97→0.98; operator confirmed the
general-tag confidence was already well tuned, so revert that. The support floor
(min_positives 30→50) and CCIP match confidence (0.92→0.95) stay. Migration 0081
(not yet deployed) edited to drop the precision bump.

Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01CDgx8bQS5YrGRK76v8HUnM

alembic/versions/0081_stricter_auto_apply_defaults.py

@@ -1,10 +1,11 @@
-"""stricter auto-apply defaults (milestone 139) — cut graduate/auto-apply misfires
+"""stricter auto-apply defaults (milestone 139) — cut auto-apply misfires
 
-head_auto_apply_precision 0.97→0.98, head_auto_apply_min_positives 30→50,
+head_auto_apply_min_positives 30→50 and ccip_auto_apply_threshold 0.92→0.95
-ccip_auto_apply_threshold 0.92→0.95 (operator-asked 2026-07-06). The model
+(operator-asked 2026-07-06). The head graduation precision bar stays 0.97 — the
-defaults change for fresh installs; here we bump the existing singleton row IFF
+operator confirmed the general-tag confidence was already well tuned; only the
-it is still at the previous default, so a deliberate operator change is NOT
+support floor + the CCIP match confidence are raised. The model defaults change
-clobbered.
+for fresh installs; here we bump the existing singleton row IFF it is still at
+the previous default, so a deliberate operator change is NOT clobbered.
 
 Revision ID: 0081
 Revises: 0080
@@ -21,10 +22,6 @@ depends_on: Union[str, Sequence[str], None] = None
 
 
 def upgrade() -> None:
-    op.execute(
-        "UPDATE ml_settings SET head_auto_apply_precision = 0.98 "
-        "WHERE head_auto_apply_precision = 0.97"
-    )
     op.execute(
         "UPDATE ml_settings SET head_auto_apply_min_positives = 50 "
         "WHERE head_auto_apply_min_positives = 30"
@@ -36,10 +33,6 @@ def upgrade() -> None:
 
 
 def downgrade() -> None:
-    op.execute(
-        "UPDATE ml_settings SET head_auto_apply_precision = 0.97 "
-        "WHERE head_auto_apply_precision = 0.98"
-    )
     op.execute(
         "UPDATE ml_settings SET head_auto_apply_min_positives = 30 "
         "WHERE head_auto_apply_min_positives = 50"

backend/app/models/ml_settings.py

@@ -51,10 +51,7 @@ class MLSettings(Base):
         Integer, nullable=False, default=8
     )
     head_auto_apply_precision: Mapped[float] = mapped_column(
-        # Stricter graduation bar (was 0.97) to cut auto-apply misfires
+        Float, nullable=False, default=0.97
-        # (operator-asked 2026-07-06): a higher precision target → fewer heads
-        # graduate and those that do get a higher per-head auto_apply_threshold.
-        Float, nullable=False, default=0.98
     )
     # Earned auto-apply (#114). A graduated head fires (tags images without a
     # human) when this master switch is on AND the head has at least