[petlenz]

Closes #106. Part of the uvwmaterial port (#94); stacked on #129.

phase (introduced here, the phase node of every composite): a phase is a bundle of produced properties, never a material. tangent_source (+ tangent_property), fraction_source (+ fraction_property, default value: a constant_scalar when fixed, "cure", "state" when it evolves — there is no parameter fallback) and, for the mean-field schemes, eshelby_source / depolarisation_source (the nodes of #128/#129, default property tensor). Outputs tangent, volume_fraction and, when wired, eshelby, depolarisation. One callback; the fraction must be non-negative. No material_ref: a composite that consumes a phase consumes its properties and the engine orders everything.

rule_of_mixture is a tangent producer: phases ("phases": ["stiff", "soft"]), kernel voigt (equal strain), reuss (equal stress) or hill (their mean). Outputs tangent and the strain concentration tensors concentration_<phase> with Σ cᵢ Aᵢ = Isym (Voigt: Isym; Reuss: Cᵢ⁻¹ : C_R; Hill: their mean, consistent with the Hill tangent). The stress is a linear_stress on the tangent, as for every tangent producer. The fractions are summed and checked per update, so phases with changing fractions or tangents (cure, damage) work. Replaces small_strain_rule_of_mixture_material and its kernel base.

Tests (test_rule_of_mixture, 8): Voigt is the arithmetic and Reuss the harmonic mean of K and G; Hill is their mean and lies between; a single phase returns its tangent; for every kernel the linear_stress gives C_eff : ε, the concentrations average to Isym and the phase stresses average to the composite stress; bad fractions and kernels throw; creation from JSON; a phase republishes its Eshelby and depolarisation tensors and a bare phase publishes none.

Report docs/materials/rule_of_mixture/report.json: the three kernels in one graph (hill with its linear_stress as the point, Voigt and Reuss recorded), checks against E_V, E_R, E_H and the ordering, the Hill concentrations averaging to one.

Also in this PR: the report generator resolves derived values in dependency order.

835/835 tests pass (GCC 14, reports on).

JSON input

New material types: phase, rule_of_mixture. A json_model document that loads (the FromJson test of the PR):

{"materials": [
  {"type": "external_strain_source", "name": "strain"},
  {"type": "constant_scalar", "name": "Ka", "value": 100.0}, {"type": "constant_scalar", "name": "Ga", "value": 40.0},
  {"type": "constant_scalar", "name": "Kb", "value": 10.0},  {"type": "constant_scalar", "name": "Gb", "value": 3.0},
  {"type": "isotropic_tangent", "name": "Ca", "K_source": "Ka", "G_source": "Ga"},
  {"type": "isotropic_tangent", "name": "Cb", "K_source": "Kb", "G_source": "Gb"},
  {"type": "constant_scalar", "name": "ca", "value": 0.3},
  {"type": "constant_scalar", "name": "cb", "value": 0.7},
  {"type": "phase", "name": "pa", "tangent_source": "Ca", "fraction_source": "ca"},
  {"type": "phase", "name": "pb", "tangent_source": "Cb", "fraction_source": "cb"},
  {"type": "rule_of_mixture", "name": "rom", "phases": ["pa", "pb"], "kernel": "voigt"},
  {"type": "linear_stress", "name": "composite", "tangent_source": "rom", "strain_source": "strain"}
]}

phase: tangent_source/tangent_property, fraction_source/fraction_property, optional eshelby_source/eshelby_property, depolarisation_source/depolarisation_property; outputs tangent, volume_fraction, eshelby, depolarisation. rule_of_mixture: phases, kernel voigt | reuss | hill; outputs tangent, concentration_<phase>.

[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.

rule_of_mixture — needs changes (minor)

• [should] compliance and per-phase inverses computed for Voigt too, and each tangent inverted twice → invert once into S[i] only when needed.
• [should] four hand-written if (k == "...") enum chains across rule_of_mixture: Voigt, Reuss and Hill effective stiffness #130/inclusion and mean_field_composite: dilute, Mori-Tanaka and self-consistent #131/short_fibre_composite: Mori-Tanaka over the fibre orientation distribution #133 → one parse_enum<E>(who, value, table) in core.
• [should] parallel vectors m_tangents/m_concentrations (and four in inclusion and mean_field_composite: dilute, Mori-Tanaka and self-consistent #131) → struct phase {…} + range-for (idiom in weighted_sum.h:115); shared effective_medium_base for strain input, stress/tangent outputs, stress = C:ε.
• [should] fractions static while inclusion publishes volume_fraction as a property → ROM of curing phases impossible; accept phases providing volume_fraction.
• [nit] 6-line header comments (all four composites); any_of/fold_left; expressions.py except NameError: continue loses the typo text; no test varies a phase tangent.

[petlenz]

Producer/consumer pass: rule_of_mixture no longer carries a fractions parameter. Its phases are inclusion nodes (now introduced here, below the mean-field PR) that produce tangent and volume_fraction; a fraction is a produced property (constant_scalar when fixed, a cure when it evolves) and the sum-to-one check runs per update on the produced values. The report builds the two phases from c_stiff/c_soft constants.