CO-EMERGENCE AS CONSTITUTIVE DYNAMICS — verification (two faces)
================================================================
Co-emergence folded into phi_pool's per-cell update as C(Ω) = C_amp + C_θ,
so the three faces are part of the tick, not an after-check. φ has two images
on a cell's state (A,θ) ∈ S¹×ℝ⁺, so C has two faces:

C_amp (amplitude face) — coemergence_force():
    u     = A/√φ ;  r_sq = u²−(u+1) ;  r_inv = u−(1+1/u)   (both 0 at φ)
    C_amp = −φ⁻¹·∇½(r_sq²+r_inv²)/√φ   bounded ±0.5, floor u≥0.25
  RESULT: kur_R 0.10→0.48, PLUCK→SUSTAIN over 120 steps. Global sync rises.

C_θ (phase face, direct-θ) — the golden angle, φ's image on S¹:
    ψ★    = 2π/φ² = 137.5077°           phyllotactic lock, NOT trivial 0-sync
    C_θ   = g·Σⱼ Aⱼ·sin(Δθᵢⱼ − ψ★)/Aᵢ  g=0.08
  RESULT (GRID=32, settle 150, genome 0x5625BA88 preserved):
    neighbour Δθ distribution shifts from 80% near 0° (amp-only, trivial sync)
    to 22.4% within ±15° of 137.5°; mean|Δθ| 20°→105°. Gain-monotone:
      g=0.02 → 3.4% near GA
      g=0.04 → 14.4%
      g=0.06 → 19.6%
      g=0.08 → 22.4%

HONEST LIMITS (stated, not hidden):
  • C_θ LOWERS kur_R by design — golden angle is anti-synchronous; global sync
    (kur_R) is the WRONG metric for phyllotactic lock. The right metric is
    clustering of neighbour Δθ at ψ★, which C_θ measurably improves.
  • Full phyllotactic lock is geometry-bounded (~22% ceiling) on a 6-neighbour
    CUBIC lattice. Clean lock needs Vogel-spiral placement, not a cube. The
    force is correct; the lattice geometry caps how tight the lock can get.

FILE MERGE: hdgl_substrate.hdgl + hdgl_boot.hdgl → hdgl.hdgl (one file, 10 glyphs).
  Boot was never separate — it is Ω advancing through coordinate τ; τ/BASIN/HOST
  are all Π(Ω). The old boot file already imported from the substrate. One
  operator T now generates substrate, computer, memory, radio, peer-sync, boot.
