Standardize student workflows and add a layered DCB teaching example

[allamaprabhuani]

Summary

• Provide a common schema-2 setup sequence for small quasi-static, explicit-dynamic, and heterogeneous-fracture teaching examples through the existing numerical runners.
• Add a layered DCB-style example with a tougher circular region, retained field images and animation, matched coarse comparison inputs, and compact data with a checksum-checked replot script.
• Make HDF5 the default requested trajectory format while preserving explicit Zarr output and existing-store compatibility.
• Add student-facing setup guidance, configuration conventions, and public documentation style guidance; update the executable teaching notebooks.
• Reject invalid multi-material properties, empty selections, conflicting constraints, and unsupported output requests before solving. Add schema-2 support to explain-config without generating a mesh.

Verification

• 467 tests passed locally on Python 3.10.
• Strict Sphinx HTML build passed with all five teaching notebooks executed and warnings treated as errors.
• The documented explanation and validation commands passed for both standard SENT exercises and all three DCB inputs.
• Both small standard exercises execute in the regression suite; HDF5 output and legacy numerical-route parity are checked.
• The retained DCB comparison replots from the committed numerical data and original triangle connectivity.
• Public payload checks excluded raw trajectories and private notes; retained CSV line endings are fixed for cross-platform checksum reproducibility.

Scientific scope

The DCB material regions are bulk AT2 fracture materials, not an independent cohesive or interphase-debonding law. Retained results show branching and different crack advance under a toughness contrast, not completed inclusion bypass or penetration. They are qualitative teaching evidence, not experimental validation or a mesh-independence claim. The historical generating source identity was not recorded and is explicitly marked unattested; no new full DCB rerun is claimed. Independent numerical and newcomer reviews informed the input-safety repairs and evidence wording.