refactor(aocscf): replace Dz0SCF with average-occupation ROKS

[wtpeter]

Summary

Replace the dz0scf reference with an average-occupation ROKS interface while retaining the existing orbital optimization and high-spin gradient definition. This PR intentionally removes the old DZ0SCF API.

Changes relative to Dz0SCF

• Construct the reference with mol.ROKS(xc=...).average_occ() (also supported on plain SymAdaptedROKS) instead of DZ0SCF(mol, xc=...). Importing nest.aocscf registers the method. The old nest.dz0scf module, DZ0SCF factory, and direct gradient aliases are removed.
• SCF still optimizes orbitals with the singly occupied orbitals shared equally between alpha and beta. The potential now comes from PySCF's RKS get_veff on the total density and is supplied to ROHF as two identical spin blocks. The old implementation called UKS get_veff with two equal spin densities.
• During SCF, energy_tot() evaluates the average-occupation energy. After _finalize, mf.e_avg_occ holds that energy and mf.e_tot holds the high-spin energy evaluated with the converged orbitals. Previously mf.e_tot held the average-occupation energy and mf.high_spin_energy() was a separate call. Consequently, mf.energy_tot() and the finalized mf.e_tot have different meanings.
• Rename the analytic gradient driver to nest.grad.aocscf. Its Z-vector equation, explicit high-spin derivative, and numerical reference tests are retained. The material implementation change is that the intermediate RKS/ROKS evaluators use lib.view(mf, ...) instead of constructing fresh objects and copying selected DFT settings. Both mf.nuc_grad_method() and mf.Gradients() return this driver.
• Split the old combined Dz0SCF/NTTDA regression tests into AOCSCF energy tests and NTTDA tests. The SVWN/B3LYP reference energies, NTTDA excitation energies, and the two gradient reference arrays and tolerances are unchanged. Add AOCSCF and NTTDA examples, with a link from examples/aocscf to the NTTDA example.

Reviewer focus

Please check the RKS versus equal-spin UKS potential equivalence, the finalized e_tot / energy_tot() contract, and whether the lib.view change preserves the intended high-spin gradient for other XC functionals and symmetry-adapted references. The gradient tests still cover NH2/PBE and triplet CH2/B3LYP; they do not cover the symmetry-adapted gradient.

Verification

• OMP_NUM_THREADS=2 python -m pytest src/nest/aocscf/tests/test_aocscf.py src/nest/grad/tests/test_aocscf_grad.py src/nest/nttda/tests/test_nttda_aocscf.py -q — 6 passed.
• Smoke check: mf.Gradients() and mf.nuc_grad_method() both return nest.grad.aocscf.Gradients.
• python -m ruff check . and git diff --check — passed.

[Copilot]

Copilot was unable to review this pull request because the user who requested the review has reached their quota limit.