wip(rls): engine-side fail-loud + harness seed identity [pin: work interrupted]

Design correction discovered by the auth chicken-and-egg: the
decorator-level refusal was re-created a convention gate (plan B
thinking). ScopedDB now NEVER refuses — it declares the GUC when the
context carries an identity and declares nothing otherwise; the ENGINE
is the fail-loud layer (policies raise on governed tables, ungoverned
tables like auth lookups simply work). rls.WithInternal removed as a
dead concept.

- search service's raw DBTX path now receives the ScopedDB itself
  (main.go + harness) — a bare pool bypassed the GUC.
- database/schema.go: migration tx declares a synthetic identity
  (schema.sql's dedup DO block SELECTs media_items; with policies
  live, boot 2 raised 'unrecognized configuration parameter' — A11).
- harness: TestServerSetup.DB is a raw-pool SEED handle (write
  policies permissive, other seeded tables ungoverned); the server
  under test keeps the scoped wiring. 33 CreateMediaItem sites use
  seedCtx() — a seed identity with no visibility rows — because
  RETURNING requires SELECT visibility of the new row.
- invariant test: regex UUID-scraping replaced with a structured JSON
  decode of the sections payload.

INTERRUPTION NOTE: an edit to this same effect (seedCtx) was applied,
confirmed by the tool, and subsequently vanished from disk while an
'opencode serve' process (PID 84551) is alive on the machine. This
commit pins the tree against further loss; implementation is PAUSED
pending the user killing that process.

internal/database/schema.go

@@ -178,6 +178,16 @@ func verifyFunctions(ctx context.Context, db *pgxpool.Pool, expectedFunctions []
 	return nil
 }
 
+// schemaMigrationIdentity is the synthetic identity the schema
+// migration transaction declares for RLS (plan appendix A11). The
+// schema's dedup DO block SELECTs/DELETEs media_items, and with the
+// visibility policies live (FORCE applies to the owner) that SELECT
+// needs app.user_id set even in this maintenance tx. The identity has
+// NO library_visibility rows, so the policy grants it full visibility
+// — appropriate for the server's own schema pass, and unreachable by
+// requests (this tx only ever runs at startup, on the raw pool).
+const schemaMigrationIdentity = "00000000-0000-0000-0000-000000000001"
+
 // executeSchema runs the entire schema.sql in a single transaction
 func executeSchema(ctx context.Context, db *pgxpool.Pool) error {
 	tx, err := db.Begin(ctx)
@@ -186,6 +196,12 @@ func executeSchema(ctx context.Context, db *pgxpool.Pool) error {
 	}
 	defer tx.Rollback(ctx)
 
+	// Declare the migration identity BEFORE the file executes (A11):
+	// policy'd tables are queried by schema.sql itself on re-runs.
+	if _, err := tx.Exec(ctx, `SELECT set_config('app.user_id', $1, true)`, schemaMigrationIdentity); err != nil {
+		return fmt.Errorf("schema execution failed (declare migration identity): %w", err)
+	}
+
 	_, err = tx.Exec(ctx, SchemaFile)
 	if err != nil {
 		log.Printf("HINT: Run manually to debug: psql -h localhost -U postgres -d bookhoard -f database/schema/schema.sql")