Add strut-curvature design-axis exploration synthesis

[Copilot]

Companion synthesis for the issue asking how strut curvature could shape the project beyond width/slenderness sweeps, including the explicit creative idea of paired struts buckling together into a bistable, locked mode. Modeled on edison-trajectories/tpu-petg-bo-variables-additions-from-pr22.md (PR #24).

Changes

• New file: edison-trajectories/strut-curvature-explorations.md

• §1 Why curvature ≠ width. Imperfection-seeded buckling sets direction of collapse independent of P_cr; turning print-induced bow into a deterministic axis converts noise into a control knob (programmed snap, lower repeatable peak F, geometric pre-stress that offsets the small TPU 85A prestress budget).

• §2 Seven new BO axes to slot alongside the 5ae24eaf table and PR Explore TPU+PLA BO design variables and base structures (Edison literature result + PR #22 design-gaps synthesis, hierarchical search space) #24 §A/§B:

  Axis	Domain	Conditional on
  strut_shape	{straight, single_arc, S_arc, helical, tapered_arc}	—
  curvature_amplitude δ/L	[0, 0.05]	shape ≠ straight
  curvature_plane φ	[0°, 90°]	arc shapes
  helix_pitch_ratio p/L	[0.5, 4]	helical
  taper_ratio	[0.6, 1.0]	tapered/helical
  pair_coordination	{independent, parallel_sign, antiparallel_sign, chiral_pair}	N_struts ≥ 2
  snap_through_target (constraint)	[0.05, 0.5]·P_cr	bistable_mode ≠ monostable

• §3 Coordinated-pair bistable mode — the issue's creative ask. Two parallel-sign curved struts latch via a printed TPU 85A bump-and-pocket at mid-span: hysteretic EA (~halved rebound vs monostable), 2^N_bay passive shape space across stacked Snelson bays, lateral squeeze re-arms.

• §4 Manufacturability on H2D. δ/L ≤ 5% stays inside the 3-perimeter envelope at strut Ø ≥ 2 mm; helical printable in-place down to p/L ≈ 0.5. Joint coupling to issue Extensively explore and validate the joint design for PETG + TPU prints to mimic tensegrity #38 is a one-line bore rotation α = atan(2δ/L) in B_dovetail.scad / A_anchor_bulb.scad — no new joint family.

• §5 Failure modes. Curvature shifts dominant mode toward hinge-yield — log per-specimen so the surrogate learns the flip. Bistable pairs add cyclic-fatigue risk on N_reuse; latched state is class-2, gate behind joint-design Phase-4.

• §6 Next actions. strut_shape arg in simulations/printable_design.py + re-run simulations/newton_drop.py; Edison LITERATURE_HIGH on coordinated-pair bistable buckling (Restrepo 2015, Frenzel 2016, Pan 2023, Schenk & Guest); conservative initial enable-set (curvature on, paired bistability gated).

No code, dependency, or CI changes.