feat(records): add duplicate detection and marking

Summary

• Adds is_duplicate (bool, indexed) and duplicate_of (UUID FK, self-referential) fields to Record model
• Detects exact duplicates at upload time via SHA-256 file hash comparison; marks new record with is_duplicate=True and duplicate_of=<original_uid>
• Adds POST /{record_id}/mark-duplicate and DELETE /{record_id}/mark-duplicate endpoints for admin/reviewer manual override
• Adds is_duplicate filter to GET /records/
• Adds Alembic migration b2c3d4e5f6a7

Test plan

• Migration b2c3d4e5f6a7 applies cleanly
• Upload original file → "is_duplicate": false, "duplicate_of": null
• Upload same file again → "is_duplicate": true, "duplicate_of": "<original_uid>"
• GET /records/?is_duplicate=true → returns only duplicate records
• GET /records/?is_duplicate=false → returns only non-duplicate records
• POST /{uid}/mark-duplicate?duplicate_of={same_uid} → 400 "A record cannot be a duplicate of itself"
• POST /{uid}/mark-duplicate?duplicate_of={other_uid} → "is_duplicate": true
• DELETE /{uid}/mark-duplicate → "is_duplicate": false, "duplicate_of": null

Checklist

• Code follows project API guidelines
• Documentation is updated (OpenAPI docs reflect new fields and endpoints)
• Code adheres to project coding standards

Closes #50