Add P_constraint=1 to solve_qmd and solve_imd for correct axial tensor behavior

[Copilot]

solve_qmd and solve_imd returned too few independent components for magnetic point groups with anti-unitary operations. For 31m (3 unitary + 3 anti-unitary ops), QMD gave 2 instead of 4, IMD gave 1 instead of 3.

QMD/IMD tensors are axial (pseudo) tensors requiring P_constraint=1 alongside T_constraint=1. The effective parity coefficient is det(Rs)^T_constraint, so for anti-unitary improper rotations (θσ_v), the T-sign (−1) and P-sign (−1) cancel, and these operations contribute the same constraints as pure spatial rotations.

Changes

• tensorconstraint.py: Added P_constraint=1 to solve_qmd() and solve_imd() calls to solve_tensor_constraints()
• test_tensorconstraint.py: Updated expected values in 5 affected tests; added test_solve_qmd_imd_magnetic_31m regression test verifying 4/3 independent components and nullspace structure

Before/After

ops = _31m_operations()  # 3 unitary + 3 anti-unitary

# Before: P_constraint=0 (polar) — anti-unitary ops kill z-components
solve_qmd(ops)  # → 2 independent components
solve_imd(ops)  # → 1 independent component

# After: P_constraint=1 (axial) — T and P signs cancel for anti-unitary improper ops
solve_qmd(ops)  # → 4 independent components (σ_xxy group, σ_xxz group, σ_zxx=σ_zyy, σ_zzz)
solve_imd(ops)  # → 3 independent components (σ_xxy group, σ_xxz=σ_zxx group, σ_zzz)

---

✨ Let Copilot coding agent set things up for you — coding agent works faster and does higher quality work when set up for your repo.