Thermal assessment invariant correction

The older creep_hot_wall_assessment.py demo used a two-dimensional deviator with trace divided by two for a plane-strain von Mises assessment. That is not the three-dimensional J2 invariant: the constitutively constrained out-of-plane stress must participate. The core standard result projection already performs this reconstruction, including thermal eigenstrain.

The demo now consumes that existing MISES field and exports it alongside displacement, temperature and stress. It uses the maximum DG0 cell-average invariant, explicitly, rather than adding another hand-written stress formula. This migration does not change equilibrium, thermal conduction or the K-R law.

Independent checks

Four affine planar tests (plane stress/strain, with/without thermal expansion) compare standard output with a full 3D Hooke stress and its principal-stress invariant. They pass at 1e-12 relative tolerance. On the actual reduced hot-wall mesh, an independent full-3D thermal strain reconstruction gives a maximum cell-average value of 554397427.2293961 Pa; the maximum pointwise difference between the two projected fields is 4.18e-7 Pa. Corrected degree-4 quadrature sampling gives 554942269.3512055 Pa, showing separately the averaging effect.

Versioned regression, not new physical validation

Quantity	Old v1	Corrected v2
Governing stress, Pa	331321497.13443035	554397427.2293957
Illustrative rupture estimate, h	31308.280996458638	2386.718875046755
Damage at 90% of illustrative rupture time	0.4376507671	0.4376507671

The unchanged final damage follows the chosen normalized endpoint; it is not independent validation of the rupture estimate. The old Golden rejected both changed quantities before its reference was explicitly versioned to q1-6x10-heat4-kr40-plane-strain-mises-2. Neither reference is an experimental material fit, and a cell-average maximum is not hotspot mesh convergence. This finding concerns the older illustrative hot-wall demo, not the separate P92 research material or the composite pressing simulation.