test: opt-in scaled_shear_n256 four-case shear diagnostic

- run_case: --diagnostic scaled_shear_n256 (requires --flow shear_wave;
  U256/U512/constant R256/projected R256, Re100, no early trace); N512
  only via one of the two scaled diagnostics; scaled_n256 stays TG-only.
- assess_shear_wave: scaled=True contract reusing 92b8afe scaled
  case/build/workload checks, shear diagnostic identity; signed rate bias,
  refined excess over each control; hash-pinned de5b0cc N128 reference and
  descriptive N128->N256 signed trend; CLI --diagnostic/--reference-files.
- tests: scaled contract, set mixing, trend/reference pin, runner options.
- shear_wave.py and oracle.py unchanged.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

tests/validation/taylor_green/test_shear_wave.py

@@ -4,8 +4,13 @@ Run from this directory with PYTHONPATH=<checkout>/src (same as test_oracle.py).
 """
 
 import hashlib
+import io
 import json
 import math
+import sys
+import tempfile
+from contextlib import redirect_stdout
+from unittest.mock import patch
 import unittest
 from pathlib import Path
 
@@ -361,7 +366,7 @@ class RunnerSampleTests(unittest.TestCase):
         self.assertEqual(row["leaf_cells"], N * N)
 
 
-def receipt(n, depth, explosion="constant"):
+def receipt(n, depth, explosion="constant", diagnostic=None):
     """Synthetic receipt satisfying every contract field of the fixed four-case set."""
     import assess_projected_pairs as tg
     steps = run_case.sample_steps(n)[0]
@@ -369,7 +374,7 @@ def receipt(n, depth, explosion="constant"):
     for row, step in zip(rows, steps):
         row.update(step=step, tstar=step * U / n,
                    spurious_pressure_scaled_by_level=[row["spurious_pressure_scaled"]] * depth)
-    cells = [n * n] if depth == 1 else [8704, 30720]
+    cells = [n * n] if depth == 1 else [17 * n * n // 32, 15 * n * n // 8]
     hops = [16 * n * 2**level for level in range(depth)]
     builds = []
     for level in range(depth):
@@ -406,7 +411,8 @@ def receipt(n, depth, explosion="constant"):
                       active_collision_updates=sum(c * h for c, h in zip(cells, hops))),
         actual_build_sidecars=builds, rows=rows, completed=True,
         flow="shear_wave", mode=list(sw.MODE), polarization=sw.E.tolist(),
-        angle_deg=sw.ANGLE_DEG, phase=sw.PHI, eps=sw.eps(n))
+        angle_deg=sw.ANGLE_DEG, phase=sw.PHI, eps=sw.eps(n),
+        **({"diagnostic": diagnostic} if diagnostic else {}))
 
 
 class ContractTests(unittest.TestCase):
@@ -448,6 +454,89 @@ class ContractTests(unittest.TestCase):
         self.assertEqual(a(self.records, "wrong-revision")["status"], "INVALID_SET")
 
 
+def scaled_set(rate_error=0.0):
+    d = run_case.SCALED_SHEAR
+    out = [receipt(256, 1, diagnostic=d), receipt(512, 1, diagnostic=d),
+           receipt(256, 2, diagnostic=d), receipt(256, 2, "projected_linear", diagnostic=d)]
+    if rate_error:
+        out[3]["rows"] = rows_for(256, rate_error=rate_error)
+        for row, step in zip(out[3]["rows"], run_case.sample_steps(256)[0]):
+            row.update(step=step, tstar=step * U / 256,
+                       spurious_pressure_scaled_by_level=[row["spurious_pressure_scaled"]] * 2)
+    return out
+
+
+class ScaledContractTests(unittest.TestCase):
+    def setUp(self):
+        import assess_shear_wave
+        self.mod = assess_shear_wave
+
+    def test_scaled_contract_and_signed_excess(self):
+        result = self.mod.assess(scaled_set(1.5e-3), "r", scaled=True)
+        self.assertTrue(result["contract_valid"], [k for k, v in result["checks"].items() if not v])
+        self.assertEqual([c[1] for c in result["cases"]], [256, 512, 256, 256])
+        projected = result["signed"]["projected_linear"]
+        self.assertAlmostEqual(projected["signed_rate_bias"], 1.5e-3, delta=1e-9)
+        self.assertAlmostEqual(projected["excess_over_uniform_root"]["signed_rate_bias"], 1.5e-3,
+                               delta=1e-9)
+        self.assertIn("excess_over_uniform_fine", projected)
+
+    def test_scaled_and_default_sets_do_not_mix(self):
+        n128 = [receipt(128, 1), receipt(256, 1), receipt(128, 2), receipt(128, 2, "projected_linear")]
+        self.assertEqual(self.mod.assess(n128, "r", scaled=True)["status"], "INVALID_SET")
+        self.assertEqual(self.mod.assess(scaled_set(), "r")["status"], "INVALID_SET")
+        unlabeled = scaled_set()
+        unlabeled[0].pop("diagnostic")
+        self.assertEqual(self.mod.assess(unlabeled, "r", scaled=True)["status"], "INVALID_SET")
+        tg_label = scaled_set()
+        tg_label[2]["diagnostic"] = run_case.SCALED_DIAGNOSTIC
+        self.assertEqual(self.mod.assess(tg_label, "r", scaled=True)["status"], "INVALID_SET")
+
+    def test_resolution_trend_and_reference_pin(self):
+        reference = [receipt(128, 1), receipt(256, 1), receipt(128, 2),
+                     receipt(128, 2, "projected_linear")]
+        new = self.mod.assess(scaled_set(1.5e-3), "r", scaled=True)
+        trend = self.mod.resolution_trend(new, reference, "r")
+        self.assertTrue(trend["valid"])
+        change = trend["changes"]["projected_linear"]["signed_rate_bias"]
+        self.assertAlmostEqual(change["signed_change"], 1.5e-3, delta=1e-9)
+        self.assertIn("excess_over_uniform_fine.residual_l2", trend["changes"]["constant"])
+        with tempfile.TemporaryDirectory() as tmp:
+            paths = []
+            for i, r in enumerate(reference):
+                paths.append(Path(tmp) / f"{i}.json")
+                paths[-1].write_text(json.dumps(r))
+            with self.assertRaises(ValueError):
+                self.mod.load_reference(paths)
+
+    def test_scaled_runner_options(self):
+        repo = Path(__file__).resolve().parents[3]
+        with tempfile.TemporaryDirectory() as tmp:
+            base = ["run_case.py", "--repo", str(repo), "--expected-revision", "f" * 40,
+                    "--output", str(Path(tmp) / "case.json"), "--dry-run", "--re", "100"]
+            shear = ["--flow", "shear_wave"]
+            scaled = ["--diagnostic", run_case.SCALED_SHEAR]
+            for extra, valid in (
+                    (["--n", "512", "--depth", "1"] + shear + scaled, True),
+                    (["--n", "256", "--depth", "2", "--explosion", "projected_linear"] + shear + scaled, True),
+                    (["--n", "512", "--depth", "1"] + shear, False),
+                    (["--n", "128", "--depth", "2"] + shear + scaled, False),
+                    (["--n", "512", "--depth", "1"] + scaled, False),  # needs shear flow
+                    (["--n", "512", "--depth", "1", "--early-pressure", "--diagnostic",
+                      run_case.SCALED_DIAGNOSTIC] + shear, False),  # mutually exclusive
+                    (["--n", "256", "--depth", "2", "--early-pressure"] + shear + scaled, False)):
+                argv = base + extra
+                with self.subTest(argv=argv), patch.object(sys, "argv", argv), \
+                     patch.object(run_case, "verify_source", return_value={}), \
+                     patch.object(run_case, "runtime_source", return_value={}), \
+                     redirect_stdout(io.StringIO()), patch("sys.stderr", io.StringIO()):
+                    if valid:
+                        self.assertEqual(run_case.main(), 0)
+                    else:
+                        with self.assertRaises(SystemExit):
+                            run_case.main()
+
+
 class DefaultPathTests(unittest.TestCase):
     def test_frozen_oracle_is_byte_identical(self):
         digest = hashlib.sha256((HERE / "oracle.py").read_bytes()).hexdigest()