test: signed interface-probe trace diagnostic (--probe-trace)

Test-only mechanism diagnostic, no pass gates. New pinned shear_probes.py:
24 fixed root-physical points on vertical/horizontal interface-normal
transects (>=8 root cells from corners), ACTIVE-leaf selection storing
level/i/j/actual centers, root-sync rho/ux/uy at root steps 0-128 and
1920-2048 (U512 every 4 lattice steps), analytic residuals at actual
centers, schedule/finite/600 kB validation. run_case: --probe-trace for
nested U128/U512/projected d3 and twin projected d2 (not with
--final-snapshot), shear_probes.py pin, probe-only steps skip metrics,
five qualification samples unchanged. shear_wave/shear_nested/shear_twin/
oracle byte-identical.

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

docs/validation/taylor_green.md

@@ -418,3 +418,39 @@ Labels:
 - Neither contrast is additive by construction.
 - The thin-band/acoustic explanation for the nested L1 pressure is a
   hypothesis, not established.
+
+### Signed interface-probe traces (opt-in `--probe-trace`, mechanism diagnostic)
+
+`shear_probes.py` (pinned) defines 24 fixed points on two interface-normal
+transects, all at least 8 root cells from any box corner:
+- V: y = 64.1, crossing the vertical faces x = 16 (L0/L1) and x = 22 (L1/L2);
+- H: x = 90.1, crossing the horizontal faces y = 24 and y = 30;
+- offsets from the outer face: −3.9, −1.9, −0.4, 0.1, 0.6, 2.1, 4.1, 5.6, 6.1,
+  6.6, 8.1, 10.1.
+
+Allowed cases:
+- nested U128, nested U512, nested projected depth 3;
+- twin projected depth 2.
+
+The option can't be combined with `--final-snapshot`.
+
+Sampling:
+- Each point is read from the ACTIVE leaf cell containing it. The receipt
+  stores the leaf level, `i`/`j`, and the actual cell center, all in
+  root-physical units (N=128).
+- Grids differ between cases, so centers are *not* collocated. Analytic
+  fields are evaluated at each actual center (`shear_probes.residuals`), and
+  comparisons across cases are between residuals at nearby but distinct
+  centers.
+- Times are root-physical steps 0–128 and 1920–2048. U512 is sampled every 4
+  lattice steps to match.
+- Each sample stores `rho`, `ux`, `uy` from the FP64 physical populations,
+  POST-collision G at root synchronization only. No substep content is
+  observed, and grid-scale alternation can alias.
+
+What's kept and enforced:
+- The five qualification samples stay unchanged.
+- No field dumps. The trace is validated for exact schedule, finiteness and
+  a 600 kB cap. If invalid it is dropped and an error is recorded, keeping
+  the scalar receipt.
+- There are no new pass gates; this is a mechanism diagnostic only.