Use NDTensors.with_auto_fermion in ITensorsExtensions and tests

[mtfishman]

Summary

Adopt the new exception-safe scope wrapper NDTensors.with_auto_fermion (added in NDTensors v0.4.25), replacing the manual ITensors.enable_auto_fermion() / disable_auto_fermion() save-restore patterns in ITensorsExtensions.map_eigvals and across four test files.

The wrapper restores the auto-fermion state via try / finally, fixing exception-safety bugs at callsites where a thrown exception would have left the flag flipped, and also fixing one logic bug in a test that always re-enabled the flag instead of restoring its prior state.

Changes

Source

• src/lib/ITensorsExtensions/src/itensor.jl::map_eigvals: replace the manual disable/enable around the eigendecomposition with a single NDTensors.with_auto_fermion(!fermionic_itensor) do … end do-block. Numerical output is bit-identical across all four (UAF × has-fermionic-subspaces) cases on the existing test suite plus a Case-3 differential test.

Tests

• test/test_opsum_to_ttn.jl ("Multiple onsite terms"): wrap with with_auto_fermion(). Also fixes a real bug in the previous save-restore: the conditional only re-enabled the flag if it was already enabled at entry, so a run with the flag originally off would leave it flipped on.
• test/test_itensorsextensions.jl ("Fermionic eigendecomp", "Fermionic map eigvals tests"): wrap each fermionic testset with with_auto_fermion().
• test/test_opsum_to_ttn_mpo_cross_check.jl: wrap "OpSum to TTN Fermions" with with_auto_fermion(). Remove a vestigial save/restore from "OpSum to TTN" (the testset never actually toggled the flag, so the captured auto_fermion_enabled and the conditional restore were dead code).
• test/test_ttn_position.jl ("ProjTTN position"): fold the if use_qns; enable; else; disable; end conditional into with_auto_fermion(use_qns) do … end.

Version

• Patch bump 0.19.0 → 0.19.1.

Out of scope

The four read-only using_auto_fermion() queries that gate logic in package source (ITensorsExtensions/itensor.jl:63, opsum_to_ttn.jl:225/382/581) are state queries, not state changes; converting them would require larger algorithm-level changes (threading a fermion-mode parameter through call signatures). Left unchanged here.

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 72.48%. Comparing base (6e3c3b3) to head (7a68a03).
⚠️ Report is 1 commits behind head on main.

Additional details and impacted files

@@            Coverage Diff             @@
##             main     #340      +/-   ##
==========================================
- Coverage   72.50%   72.48%   -0.02%     
==========================================
  Files          67       67              
  Lines        3179     3177       -2     
==========================================
- Hits         2305     2303       -2     
  Misses        874      874              

Flag	Coverage Δ
docs	46.24% <0.00%> (-0.01%)	⬇️

Flags with carried forward coverage won't be shown. Click here to find out more.

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.