feat(catalog): real catalog display names

[mrosseel]

What

Make a readable catalog name a first-class data field so the UI can show e.g. Collinder 24 sourced from data instead of a hardcoded code→name map. Today the catalogs table only has catalog_code, max_sequence, and a long free-text desc, so the UI falls back to codes (Col 24, H 244).

Changes

• Schema (db/objects_db.py): add a name TEXT column to the catalogs CREATE TABLE and to ObjectsDatabase.insert_catalog().
• Write path: catalog_imports/catalog_import_utils.insert_catalog() gains a display_name param. All ~21 loader call sites now pass a readable name (NGC, IC, Messier, Caldwell, Herschel, Collinder, TAAS 200, SAC Asterisms/Doubles/Red Stars, Bright Stars, Extragalactic Globular Clusters, RASC Double Stars, Barnard, Sharpless, Abell, Arp, TLK Variable Stars, Washington Double Star, Harris, Lynga).
• Read path: expose name on the in-memory Catalog / CatalogBase, populated from get_catalogs_dict(). Falls back to None (UI uses the code) for virtual catalogs (planets, comets) and legacy rows.
• Migration: the shipped astro_data/pifinder_objects.db is a pre-built binary without this column. A one-time, idempotent migration runs on startup (ObjectsDatabase.__init__): if name is missing it does ALTER TABLE catalogs ADD COLUMN name TEXT then UPDATEs each row by catalog_code from a canonical map, so existing installs get names without a full re-import. The migration is race-safe (catches duplicate-column on concurrent startup) and a no-op once applied.
• Cache: bump CACHE_VERSION to 2 since the cached Catalog / catalogs_info shape changed.

The shipped binary DB is intentionally not modified in this PR — names are applied at runtime by the migration (verified: all 21 catalogs backfill correctly).

Consumer note

Step 5 of the original task was a one-line change in object_details._other_catalog_descriptions to build the section label from the catalog name (falling back to the code). That method is part of an as-yet-unmerged observing-list feature and does not exist on brickbots/main, so it is not included here. Once it lands, the change is simply: look up the catalog name via catalogs.get_catalog_by_code(code).name and render f"{name} {sequence}", falling back to the code when name is None.

Testing

• ruff check + ruff format --check: clean.
• pytest tests/test_catalog_cache.py tests/test_catalog_data.py: 11 passed.
• Verified the startup migration backfills all 21 catalog names on the real shipped DB.

🤖 Generated with Claude Code