feat: complete tensor-network API update

[rezaquant]

Summary

• complete the public tensor-network API and documentation/skill updates
• finish TreeOptimizer native fermionic paths, mutation-safe norm caching, and gauge-neutral measurements
• standardize supported TreeOptimizer modes around auto, direct, and mpo
• improve norm-infidelity progress reporting and block-aware truncation diagnostics
• update symmetric-tensor and sweep backend behavior with focused regression coverage
• remove stale benchmark/example artifacts covered by the API refresh

Validation

• pytest -q tests/test_optimize_tree.py tests/test_tree_sampler.py tests/test_tree_symmetric.py tests/test_sweep_symmray_backend.py — 226 passed
• pytest -q tests/test_public_api.py tests/test_package_layout.py tests/test_symmetric_tensors.py tests/test_trajectory_noise.py -k 'not benchmark' — 848 passed, 1 deselected
• pyflakes, py_compile, and staged diff checks passed