fix(review): queue only ambiguous files that nothing links, and heal the queue

The maintainer's live scan (19,908 files, scope "new") reported 4,677 files
under Needs review, but the queue held 10,232 items: every file the strict
resolver found ambiguous was queued, including the ones the legacy
browser-parser fallback then linked (all sampled "added" rows) and the ones
already linked to a known prompt (135 of 200 sampled "known" rows).

A file is now queued only when its final outcome is ambiguous and unlinked:
the resolver finds it ambiguous, the fallback links nothing and no saved row
links it already. An ambiguous file that a saved row links is reported as
unchanged ("Already linked to its prompt"), not as Needs review.

The queue heals itself. After each batch, the scan removes the pending items
of every listed file (read, or skipped as known) that a saved row now links.
The monitor does the same for each ambiguous output it links, and never
queues one. Only an undecided item is removed (pending, revision 0, no
decision row), so a decision is never deleted or changed and the schema
needs no new state.

The scan job's scope.review_items now counts this scan's files that are in
the pending queue after it, new or already there, so it agrees with the
review group's ambiguous rows and with the queue total. review_added and
review_settled are new, and the log gains "already in Needs review" and
"left Needs review · now linked" lines. A duplicate whose item is decided is
reported as "decided" and is not counted.

database/review.py

@@ -32,6 +32,7 @@ from .annotation_schema import (
 from .annotations import _new_id, _validate_locator, _validate_target
 from .models import normalize_image_path, utc_now_iso
 from .review_schema import (
+    DECISIONS_TABLE,
     ITEMS_TABLE,
     MAX_CANDIDATES,
     MAX_SPELLED_PATH_BYTES,
@@ -80,6 +81,26 @@ _QUEUED = (
     "WHERE f.locator=? AND b.size=? AND b.mtime_ns=? AND i.resolver_version=? "
     "LIMIT 1"
 )
+_PENDING_AT = (
+    f"SELECT 1 FROM {FILES_TABLE} f "  # nosec B608
+    f"JOIN {BYTES_TABLE} b ON b.file_id=f.id "
+    f"JOIN {ITEMS_TABLE} i ON i.byte_id=b.id "
+    "WHERE f.locator=? AND b.size=? AND b.mtime_ns=? AND i.resolver_version=? "
+    "AND i.state='pending' LIMIT 1"
+)
+# Pending items of one file: the candidates a link settles.
+_PENDING_OF_LOCATOR = (
+    f"SELECT i.id FROM {FILES_TABLE} f "  # nosec B608
+    f"JOIN {BYTES_TABLE} b ON b.file_id=f.id "
+    f"JOIN {ITEMS_TABLE} i ON i.byte_id=b.id "
+    "WHERE f.locator=? AND i.state='pending'"
+)
+# Only an undecided item goes: a decision row (RESTRICT) is never deleted.
+_SETTLE_ITEM = (
+    f"DELETE FROM {ITEMS_TABLE} WHERE id=? AND state='pending' "  # nosec B608
+    f"AND revision=0 AND NOT EXISTS (SELECT 1 FROM {DECISIONS_TABLE} d "
+    f"WHERE d.item_id={ITEMS_TABLE}.id)"
+)
 _EPOCH = "SELECT storage_epoch FROM execution_history_metadata WHERE singleton_id=1"
 _PENDING_COUNT = (
     f"SELECT COUNT(*) FROM {ITEMS_TABLE} WHERE state='pending'"  # nosec B608
@@ -342,6 +363,60 @@ class ReviewOperations:
         )
         return row is not None
 
+    def review_pending_at(
+        self, locator: str, size: int, mtime_ns: int, resolver_version: int
+    ) -> bool:
+        """True when a *pending* item cites bytes verified at this exact stat.
+
+        ``review_fingerprint_queued`` also matches decided items; a file whose
+        only item is decided is not in the queue.
+        """
+        row = (
+            self.model.get_connection()
+            .execute(_PENDING_AT, (locator, size, mtime_ns, resolver_version))
+            .fetchone()
+        )
+        return row is not None
+
+    def review_file_linked(self, spellings) -> bool:
+        """True when a saved row links the file under any of ``spellings``."""
+        with self._annotation_read() as conn:
+            return bool(dependent_images(conn, spellings))
+
+    def settle_linked_review_items(self, files) -> int:
+        """Remove the pending, undecided items of files that are now linked.
+
+        ``files`` holds ``(locator, spellings)`` per file. An item leaves the
+        queue only when it is still ``pending`` at revision 0 with no decision
+        row and a saved row links its file under one of ``spellings``; a
+        decided item is never touched. Files without a pending item cost one
+        indexed read and open no write transaction. Returns how many items
+        were removed (the review index revision moves when any was).
+        """
+        files = [(locator, tuple(spellings)) for locator, spellings in files]
+        for locator, _spellings in files:
+            _validate_locator(locator)
+        conn = self.model.get_connection()
+        candidates = [
+            (locator, spellings)
+            for locator, spellings in files
+            if conn.execute(_PENDING_OF_LOCATOR, (locator,)).fetchone() is not None
+        ]
+        if not candidates:
+            return 0
+        removed = 0
+        with self._pair_transaction() as conn:
+            for locator, spellings in candidates:
+                if not dependent_images(conn, spellings):
+                    continue
+                for (item_id,) in conn.execute(
+                    _PENDING_OF_LOCATOR, (locator,)
+                ).fetchall():
+                    removed += conn.execute(_SETTLE_ITEM, (item_id,)).rowcount
+            if removed:
+                conn.execute(_BUMP_REVIEW_INDEX)
+        return removed
+
     def review_queue_page(self, *, limit: int, after: tuple | None = None) -> dict:
         """One page of pending items, oldest first, from one snapshot.