feat: Deliver validated initial CIF preparation and workflow guidance

[D-sudoasd]

This change makes alloy starting CIF preparation usable from the CLI and desktop. Materials Project prototypes previously arrived as P1 files that the adaptation step refused; users also had to supply symmetry-equivalent cell axes manually.

The new prepare-cifs workflow validates the actual single metal orbit, restores chemically supported P1 symmetry into a conventional cell, and creates a fresh verified package containing initial CIFs, unchanged sources, per-phase provenance, a theoretical peak preview, and a refinement guide. It supports cited phase-specific lattice, coordinates and chemistry, attributed offline Ti scaffolds, host selection by atomic fraction, and configurable candidate attempts. The desktop dialog can load the result into local analysis and remains usable on small screens. Standalone adaptation links equivalent cell axes, recomputes formula/Z/mass, records output hashes and prevents concurrent replacement; provider errors redact credentials.

Current project instructions, documentation navigation and contribution/release guidance are included with the linked README changes. The ordinary analysis and discovery workflows retain their existing scientific meaning. Inherited lattice parameters and nominal bulk chemistry are explicitly starting assumptions; no experimental phase identification, equilibrium partitioning or sample refinement is inferred.

Validation on Windows / Python 3.12:

• Focused preparation, adaptation, CLI and GUI snapshot: 84 passed. The closeout rerun had 83 passed and one transient Tk-resource skip; that exact small-screen GUI test then passed separately, and fresh Tk startup succeeded.
• Full suite snapshot: 746 passed, 6 skipped for unavailable inherited CIF fixtures or Windows symlink privileges; affected modules were retested after the final scoped fixes.
• Analytic scientific benchmark: 45/45 checks passed.
• Ruff, documentation links, source compilation and Git whitespace checks.
• Actual Materials Project Ti prototypes, independent pymatgen reading of mixed-occupancy models, independent Bragg/q checks, and source/prototype/output SHA-256 checks.
• Installed wheel produces three offline Ti starting CIFs with a valid manifest; actual Tk preparation/loading and a 768-pixel small-screen layout were exercised.

The 6 environment-dependent skipped checks are not treated as experimental acceptance. Example outputs, credentials, original user data, build products and the local validation receipts are excluded from this change.