v0.0.10

What's Changed

Added

• Unified GroverResult type — All facade operations (write_files, read, edit, delete, move_files, copy_files, read_files, write_chunks, reconcile) now return GroverResult with typed Detail objects providing per-file success/failure tracking, messages, and metadata.
• Batch move_files(), copy_files(), read_files() — New facade methods for bulk operations with per-file result tracking.
• ValidatedSQLModel base class — Runtime Pydantic validation for SQLModel instances constructed from non-database sources (DataFrames, dicts). DB models opt in via ValidatedSQLModel inheritance; model_from_mount() on GroverContext resolves the correct model class per mount.
• move --follow support — move_files() gains a follow parameter to update graph connections after moves.
• list_dir returns candidates — Directory listing now returns typed result objects.
• ReconcileDetail — New detail type for per-file reconcile tracking (added/updated/deleted).

Changed

• Version/trash facade methods removed — versions(), rollback(), trash(), restore(), empty_trash() removed from facade, backends, and protocols. Deletions now cascade automatically through the database.
• read() simplified — offset and limit parameters removed from read() across facade, backends, and utilities. Content slicing is no longer a backend concern.
• write_chunks refactored — Follows the same thin-facade pattern as write_files, with chunking logic pushed into ChunkProvider.
• Ref internals unified — Internal ref parsing consolidated into models/internal/ref.py with enhanced decomposition.
• Mount naming and DB model base classes simplified — Cleaner mount initialization, streamlined dialect helpers.

Fixed

• ty 0.0.23 compatibility — Protocol mismatches and or-pattern union errors resolved for latest type checker.
• GroverResult migration alignment — Backend return types, batch type exports, and test expectations aligned with unified result model.

Full Changelog: v0.0.9...v0.0.10