fix(test): replay full migration chain in source-registry realinfra test

tests/p0/source-registry-postgres.realinfra.test.ts only applied
sql/002_source_registry.sql before running assertions. sql/007_ingestion_plane.sql
later added opt_sources.owner_id, and src/repositories/postgres-source-repository.ts's
INSERT has named that column ever since -- but nothing caught the drift because
SOURCE_REGISTRY_TEST_DATABASE_URL is never set in CI (.gitea/workflows/ci.yml has no
job that sets it), so this suite has silently skipped in every CI run.

Verified against a real disposable Postgres 16 container: before this fix, all 7
tests failed with `column "owner_id" of relation "opt_sources" does not exist`;
after replaying every sql/*.sql file in order (same pattern already used by
tests/analytics/analytics-domain-postgres.realinfra.test.ts), all 7 pass.

Found while auditing 0pt's Analytics/BI layer (MetricFlow/Cube/ClickHouse/
Trino/DuckDB) and standing up real Postgres/ClickHouse containers to verify
those claims for real; this bug is adjacent (source registry, not analytics)
but was directly encountered and is a small, safe, test-only fix.

Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>

tests/p0/source-registry-postgres.realinfra.test.ts

@@ -24,10 +24,24 @@
  * prove the rows are really there, not an artifact of the service's own
  * in-process bookkeeping.
  *
- * beforeAll applies sql/002_source_registry.sql itself (every statement is
- * `CREATE TABLE/INDEX IF NOT EXISTS`, so this is idempotent and safe to run
- * against an already-migrated database too) -- this test does not require
- * the operator to have pre-run migrations.
+ * beforeAll applies every sql/NNN_*.sql file in order (same restore-drill
+ * pattern as tests/analytics/analytics-domain-postgres.realinfra.test.ts),
+ * not just sql/002_source_registry.sql -- this test does not require the
+ * operator to have pre-run migrations.
+ *
+ * CORRECTED (this pass, 2026-09-14): it used to apply ONLY
+ * sql/002_source_registry.sql. That was enough the day this file was
+ * written, but sql/007_ingestion_plane.sql later added `opt_sources.owner_id`
+ * (`ALTER TABLE opt_sources ADD COLUMN IF NOT EXISTS owner_id TEXT`), and
+ * src/repositories/postgres-source-repository.ts's INSERT has named that
+ * column ever since. Because this env var is never set in CI (confirmed by
+ * grep over .gitea/workflows/ci.yml — no job references
+ * SOURCE_REGISTRY_TEST_DATABASE_URL), the drift was never caught: every
+ * assertion below failed with `column "owner_id" of relation "opt_sources"
+ * does not exist` the moment it was actually run against a real, freshly
+ * migrated-to-002-only Postgres. The fix is to stop hand-picking one
+ * migration file and replay the whole chain, exactly like the analytics
+ * suite already does.
  */
 import { afterAll, beforeAll, describe, expect, it } from "vitest";
 import fs from "node:fs";
@@ -56,9 +70,23 @@ beforeAll(async () => {
       adminSql`SELECT 1`,
       new Promise((_, reject) => setTimeout(() => reject(new Error("connect timeout")), 5000)),
     ]);
-    const migrationPath = path.join(process.cwd(), "sql", "002_source_registry.sql");
-    const migration = fs.readFileSync(migrationPath, "utf8");
-    await adminSql.unsafe(migration);
+    // Replay every migration in order, not just 002 -- see the file header
+    // "CORRECTED" note. Tolerate an individual file failing (mirrors
+    // analytics-domain-postgres.realinfra.test.ts: sql/001 needs the
+    // `vector` extension, which a stock postgres:16-alpine does not carry)
+    // as long as the files this test actually depends on (002, 007) apply.
+    const dir = path.join(process.cwd(), "sql");
+    for (const file of fs
+      .readdirSync(dir)
+      .filter((f) => f.endsWith(".sql"))
+      .sort()) {
+      const body = fs.readFileSync(path.join(dir, file), "utf8");
+      try {
+        await adminSql.unsafe(body);
+      } catch (err) {
+        console.warn(`[realinfra] ${file} did not apply cleanly: ${String(err).slice(0, 200)}`);
+      }
+    }
     reachable = true;
   } catch (err) {
     reachable = false;