docs: citation audit + correct 8 mis-citations; add Phases 12/13

External reviewer pass over the literature references. Verified entries
against arXiv/DOI/publisher and corrected misattributions that had
propagated across the docs.

Corrected citations (consistent across all docs):
- Bowers-Bowers-Lutz 2026: title was the 2017 paper's
  -> "Rigidity of Koebe Polyhedra and Inversive Distance Circle Packings"
- Liouville theorem: "Springborn 2019" -> Pinkall & Springborn,
  Geom. Dedicata 214 (2021)
- Bobenko-Pinkall-Springborn: "G&T 14 (2010)" -> G&T 19(4) (2015), 2155-2215
- Optimal Cone Singularities: "Crane, Soliman, Ben-Chen, Schroeder"
  -> Soliman, Slepcev, Crane, ACM TOG 37(4)
- Schlaefli formula: "Rivin, Springborn 1999" -> Rivin, Schlenker
- Quasiconformal distortion: "Springborn, Veselov" -> Born, Buecking,
  Springborn (arXiv:1505.01341)
- Period matrices: "Bobenko, Buecking 2009" (was Bobenko-Mercat-Schmies'
  title) -> Bobenko-Mercat-Schmies 2011 + genuine Bobenko-Buecking 2021
- Fabricated entry: "Alexa 2020, DOI 10.1145/3414685.3417840" pointed to
  an unrelated paper (Pixelor) -> Bunge, Herholz, Kazhdan, Botsch 2020
- Stripe Patterns: "Bonneel et al. 2015" -> Knoeppel, Crane, Pinkall,
  Schroeder 2015

Equation-number corrections (verified against the PDFs):
- Glickenstein 2011 "eq. 4.6" -> "§5.2" (no such equation label exists)
- Springborn 2020 "eq. 4.6" -> "§4 variational gradient"
- inversive-distance attribution softened to classical inversive distance

Other:
- DBFEnergy bibliography (separate repo) and convergence half-sentence in
  novelty-statement.md §3.3 (Bobenko-Buecking 2021)
- Status legend (implemented vs planned) at top of references.md
- New Phase 12 (decorated DCE & geometric transition, Chain A, near-term)
  and Phase 13 (canonical tessellations & polyhedral realisation, Chain B
  capstone) in phases.md + research-track.md; 10c scope-boundary note
  clarifying infrastructure vs Lutz-specific algorithms

Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>

CLAUDE.md

@@ -156,7 +156,7 @@ Hessian sign and solver per model:
 - **Spherical:** H is NSD (concave energy) → `SimplicialLDLT(−H)` (sign flip inside `newton_spherical`).
 - **HyperIdeal:** H is PSD (strictly convex) → `SimplicialLDLT(H)`.  Phase 9b uses a **block-FD Hessian** (per-face 6×6 local block, ~96× speed-up vs full FD on V=200).  Full analytic Hessian via the chain `(bᵢ, aₑ) → lᵢⱼ → ζ₁₃/ζ₁₄/ζ₁₅ → αᵢⱼ/βᵢ` is planned research — see `doc/roadmap/research-track.md` Phase 9b-analytic.
 - **CP-Euclidean:** analytic 2×2-per-edge `h_jk = sin θ / (cosh Δρ − cos θ)` (BPS 2010), strictly convex → `SimplicialLDLT(H)`.
-- **Inversive-Distance:** FD Hessian (inline in `newton_inversive_distance`).  Analytic via Glickenstein 2011 eq. (4.6) is planned research (Phase 9a.2-analytic).
+- **Inversive-Distance:** FD Hessian (inline in `newton_inversive_distance`).  Analytic via Glickenstein 2011 §5.2 is planned research (Phase 9a.2-analytic).
 
 When `SimplicialLDLT` fails (rank-deficient H — gauge mode on a closed mesh without pinned vertex/face), the solver automatically retries with `Eigen::SparseQR` to find the minimum-norm step orthogonal to the null space.  Public API: `solve_linear_system(H, rhs, &used_fallback)`.