test(tier3): add 9 SIM_verif_attach_mass RUNs (#99 Bucket B)

[simnaut]

Summary

Closes part of #99 Bucket B by extending the structural mass-tree
cross-validation in tier3_sim_attach_mass.rs from 8 → 17 of JEOD's 21
SIM_verif_attach_mass RUNs.

This is a structural mass-tree validation, not a trajectory test: JEOD's
sim is initialization-only (Trick stops at t=0, or t=2s for runtime
detach/reattach), dumping the composite tree via MassBody::print_tree(). Each
new RUN gets a build_run_*() that constructs the same MassTree from the
JEOD Modified_data/*.py + input.py config, and we compare composite
mass / CoM / inertia against the committed .out fixtures.

RUNs added (9)

Offset-attach group

• RUN_05 — single parent, Struct inertia spec, non-zero CoM [0.5,0,1] (no attach)
• RUN_06 — parent (StructCG) + child1 (Spec), offset attach [0,0,2]
• RUN_07 — parent (StructCG) + child1 (SpecCG), offset attach [-1,0,0]

Named-point group (BodyAttachAligned)

• RUN_103 — 3-body chain via named points (inline Euler orientations)
• RUN_104 — 3-body chain via named points (InputMatrix orientations)
• RUN_106 — parent (StructCG) + child1 (Spec) via named points
• RUN_107 — parent (StructCG) + child1 (SpecCG) via named points
• RUN_110 — named-point attach of 3 children, then runtime detach of child2
• RUN_111 — named-point + offset attach, then runtime reattach of child2

New physics ported (config-only)

The Spec and SpecCG inertia_spec branches of
mass_properties_init.cc are ported as init-time transforms:
I_body = T_io · (I_user − pmi(m, inertia_offset)) · T_ioᵀ for Spec; the
offset term is dropped for SpecCG. These are pure configuration transforms
(JEOD reference data is used only for test comparison). As a
structurally-independent cross-check, both reduce to the same box inertia
diag(0.3333, 0.4167, 0.0833) that the Body spec declares directly — exactly
the JEOD test designers' intent.

Tolerances

Observed max errors across all 17 RUNs: mass 0, CoM 1.0e-16 m,
inertia 5.4e-7 kg·m². Source-code tolerances stay at the existing 5e-6
floor, which is the JEOD %20lf (6-decimal) print_tree print-precision
floor and the dominant noise source (per the CLAUDE.md 1.05×-observed-max
convention; the print floor bounds the observed numeric max with margin).

The frozen baselines.json regression snapshot is refrozen for this entry:
composite_inertia 4.0e-7 → 5.40e-7 and composite_com 4.08e-17 → 1.02e-16,
reflecting the new RUNs' observed maxima. The widening ratio is 1.35×
(under the 1.50× warn band) and composite_com sits below the 1e-12
absolute floor, so check_baseline_widening stays silent.

Not added (blockers)

• RUN_08 / RUN_108 — a child body is attached to two different parents in
  separate body actions (e.g. child3 → child2 via attach2 and child3 →
  parent via attach3). Determining the resolved final topology would require
  reading JEOD's output, not source config — so this is deferred rather than
  guessed.
• RUN_09 / RUN_109 — the root parent has a non-identity structure→body
  transform (parent_mass_orientation_optionA). JEOD reports composite
  CoM/inertia in the parent body frame (mass_update.cc:101/107 rotate the
  struct-frame offsets by composite_properties.T_parent_this), whereas our
  MassTree::recompute_composites keeps composites in the struct frame.
  Closing these needs the body-frame composite rotation ported into
  recompute_composites — flagged for a follow-up rather than approximated.

Verification

• cargo nextest run -p astrodyn_verif_jeod -E 'test(attach_mass)' — pass (17 RUNs)
• cargo run -p astrodyn_verif_jeod --bin tier3_baseline_diff — pass (refrozen entry)
• cargo fmt --check — clean
• cargo clippy -p astrodyn_verif_jeod --tests -- -D warnings — clean
• parity_coverage — pass (attach_mass remains a documented permanent gap:
  structural test, no trajectory CSV)

🤖 Generated with Claude Code

[Copilot]

Pull request overview

This PR extends JEOD SIM_verif_attach_mass structural mass-tree validation by adding new run builders, committed mass.out fixtures, and reference-generation script entries. The implementation currently covers 17 RUNs, while the PR text and module docs claim 18.

Changes:

• Adds Spec / SpecCG inertia-spec helper transforms and new build_run_* mass-tree scenarios.
• Adds JEOD print_tree fixtures for the new attach-mass cases.
• Updates the Trick reference-generation script to produce the new fixtures.

Reviewed changes

Copilot reviewed 11 out of 11 changed files in this pull request and generated 1 comment.

Show a summary per file

File	Description
crates/astrodyn_verif_jeod/tests/tier3_sim_attach_mass.rs	Adds new mass-property helpers, run builders, and validation calls.
trick/generate_references.sh	Adds new SIM_verif_attach_mass reference-generation entries.
crates/astrodyn_verif_jeod/test_data/attach_mass_05_mass.out	Adds JEOD fixture for RUN_05.
crates/astrodyn_verif_jeod/test_data/attach_mass_06_mass.out	Adds JEOD fixture for RUN_06.
crates/astrodyn_verif_jeod/test_data/attach_mass_07_mass.out	Adds JEOD fixture for RUN_07.
crates/astrodyn_verif_jeod/test_data/attach_mass_103_mass.out	Adds JEOD fixture for RUN_103.
crates/astrodyn_verif_jeod/test_data/attach_mass_104_mass.out	Adds JEOD fixture for RUN_104.
crates/astrodyn_verif_jeod/test_data/attach_mass_106_mass.out	Adds JEOD fixture for RUN_106.
crates/astrodyn_verif_jeod/test_data/attach_mass_107_mass.out	Adds JEOD fixture for RUN_107.
crates/astrodyn_verif_jeod/test_data/attach_mass_110_mass.out	Adds JEOD fixture for RUN_110.
crates/astrodyn_verif_jeod/test_data/attach_mass_111_mass.out	Adds JEOD fixture for RUN_111.

---

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.