[petlenz]

Closes #100, closes #101, closes #102. Part of the uvwmaterial port (#94); stacked on #128.

• core/isotropize.h: isotropize<T, Dim>(C) → {K, G} of the closest isotropic tensor, K = Pvol::C / Dim, G = Pdev::C / (2 (Dim(Dim+1)/2 − 1)) (the formulas of tensor_isotropization_general, written for any dimension), and isotropic_part(C). Tests: an isotropic C is its own projection; a deviatoric rank-one term shifts only G, by the predicted amount; the moduli are rotation invariant for an anisotropic C.
• core/lebedev.h with core/lebedev_tables.h generated by tools/extract_lebedev.py from uvwmaterial's sphere_lebedev_rule (Lebedev–Laikov): precisions 3, 5, 7, 9, 11, 13, 15, 23 (6 to 194 points) as unit vectors and weights summing to one, instead of the 13.9k-line table. rule(p), average(p, f) and integrate(p, f) for scalar or tensor-valued f(n). Tests: weights sum to one and nodes are unit; the second moment is I/3 and the fourth (I⊗I + 2 Isym)/15 for every precision; the degree of exactness (the 6-point rule misses n₁⁴, the 14-point rule gets it).
• eshelby::depolarisation(a1, a2, a3): N_ii = (a₁a₂a₃/3) R_J(a₁², a₂², a₃², a_i²); an infinite axis gives the elliptic cylinder N_i = a_j/(a_i + a_j); depolarisation_cylinder(direction). The original's elliptic-cylinder formula (a spheroid expression) is not carried over. Tests: sphere I/3, needle and disc limits, trace one, the elliptic cylinder, prolate and oblate spheroids against the closed forms (Osborn).

818/818 tests pass (GCC 14, reports on).

JSON input

New material types: depolarisation_sphere, depolarisation_cylinder, depolarisation_spheroid, depolarisation_ellipsoid. A json_model document that loads (the FromJson test of the PR):

{"materials": [
  {"type": "depolarisation_sphere", "name": "sphere"},
  {"type": "depolarisation_cylinder", "name": "cylinder", "direction": [0, 1, 0]},
  {"type": "depolarisation_spheroid", "name": "spheroid", "aspect_ratio": 5, "direction": [1, 0, 0]},
  {"type": "depolarisation_ellipsoid", "name": "ellipsoid", "semi_axes": [1, 2, 3]}
]}

Producers without inputs (the conductivity Eshelby tensor of an isotropic matrix is geometry only); output tensor (rank 2, trace one), and nothing else — no interface, composites consume the property. A phase (#130) republishes it through "depolarisation_source": "sphere" for the conductivity composites (#110).

[petlenz]

Review findings (design, tmech expressions instead of component loops, modern C++, comments, tests, correctness). Severity: must = wrong result, hang, build break or unreachable feature; should = design, duplication, accuracy; nit = style.

isotropize / Lebedev / depolarisation — needs changes (small)

• [should] 2D isotropisation correct but untested → isotropize<T,2>(isotropic_stiffness<T,2>(K,G)) round trip.
• [should] lebedev.h:42-48 decltype(f(n)) with a callable returning an unevaluated expression references the reused n → static_assert scalar or tmech::tensor value.
• [should] precision list exists four times → generator emits one constexpr table; rule() = ranges::find_if.
• [should] lebedev_tables.h lacks citation (Lebedev & Laikov 1999), licence note and generating command.
• [should] depolarisation uses rj(x,y,z,z) (= rd, several× cost) and recomputes the I integrals ellipsoid already has → shared detail::ellipsoid_integrals(a).
• [nit] 6.123e-17 for exact zeros, 15-digit scatter in identical values; dev_rank misnamed; -ize/-isation mix; extract_lebedev.py validation by assert.