fix: sandbox upgrade test uses 0.4.0 target (not current default)

The upgrade test previously used 0.3.0 as the target, which matched
the new spec_version default. The upgrader correctly skips no-op upgrades,
causing the test assertion 'Upgraded' in output to fail.

Fix: upgrade to 0.4.0 (a future version) so the upgrade path runs.
This is the correct test pattern — always test with a version AHEAD of
the scaffold default.

Co-Authored-By: Oz <oz-agent@warp.dev>

Author: tbitcs

tests/sandbox/test_sandbox_new.py

@@ -154,15 +154,17 @@ def test_full_scaffold_workflow(self, tmp_path: Path) -> None:
         assert compress_result.exit_code == 0
         assert "No compression needed" in compress_result.output
 
-        # ---- Step 13: Upgrade to new version ----
+        # ---- Step 13: Upgrade to a future version ----
+        # Use a version ahead of the current default so the upgrader actually runs.
+        # (If target == current spec_version, upgrader skips — that's correct behavior.)
         upgrade_result = runner.invoke(
-            main, ["upgrade", "--project-dir", str(project), "--spec-version", "0.3.0"]
+            main, ["upgrade", "--project-dir", str(project), "--spec-version", "0.4.0"]
         )
         assert upgrade_result.exit_code == 0, f"Upgrade failed: {upgrade_result.output}"
         assert "Upgraded" in upgrade_result.output
         with open(saved) as f:
             upgraded_cfg = yaml.safe_load(f)
-        assert upgraded_cfg["spec_version"] == "0.3.0"
+        assert upgraded_cfg["spec_version"] == "0.4.0"
 
         # ---- Step 14: Diff (may show differences after upgrade) ----
         diff_result = runner.invoke(main, ["diff", "--project-dir", str(project)])