Add storage abstraction, stats endpoints, garbage collection, and test infrastructure

- Add StorageBackend protocol for backend-agnostic storage interface
- Add health check with storage and database connectivity verification
- Add garbage collection endpoints for orphaned artifacts (ref_count=0)
- Add deduplication statistics endpoints (/api/v1/stats, /stats/storage, /stats/deduplication)
- Add per-project statistics endpoint
- Add verify_integrity method for post-upload hash validation
- Set up pytest infrastructure with mock S3 client
- Add unit tests for hash calculation and duplicate detection

backend/app/schemas.py

@@ -387,3 +387,72 @@ class PresignedUrlResponse(BaseModel):
 class HealthResponse(BaseModel):
     status: str
     version: str = "1.0.0"
+    storage_healthy: Optional[bool] = None
+    database_healthy: Optional[bool] = None
+
+
+# Garbage collection schemas
+class GarbageCollectionResponse(BaseModel):
+    """Response from garbage collection operation"""
+
+    artifacts_deleted: int
+    bytes_freed: int
+    artifact_ids: List[str]
+    dry_run: bool
+
+
+class OrphanedArtifactResponse(BaseModel):
+    """Information about an orphaned artifact"""
+
+    id: str
+    size: int
+    created_at: datetime
+    created_by: str
+    original_name: Optional[str]
+
+
+# Storage statistics schemas
+class StorageStatsResponse(BaseModel):
+    """Global storage statistics"""
+
+    total_artifacts: int
+    total_size_bytes: int
+    unique_artifacts: int  # Artifacts with ref_count > 0
+    orphaned_artifacts: int  # Artifacts with ref_count = 0
+    orphaned_size_bytes: int
+    total_uploads: int
+    deduplicated_uploads: int
+    deduplication_ratio: (
+        float  # total_uploads / unique_artifacts (if > 1, deduplication is working)
+    )
+    storage_saved_bytes: int  # Bytes saved through deduplication
+
+
+class DeduplicationStatsResponse(BaseModel):
+    """Deduplication effectiveness statistics"""
+
+    total_logical_bytes: (
+        int  # Sum of all upload sizes (what would be stored without dedup)
+    )
+    total_physical_bytes: int  # Actual storage used
+    bytes_saved: int
+    savings_percentage: float
+    total_uploads: int
+    unique_artifacts: int
+    duplicate_uploads: int
+    average_ref_count: float
+    max_ref_count: int
+    most_referenced_artifacts: List[Dict[str, Any]]  # Top N most referenced
+
+
+class ProjectStatsResponse(BaseModel):
+    """Per-project statistics"""
+
+    project_id: str
+    project_name: str
+    package_count: int
+    tag_count: int
+    artifact_count: int
+    total_size_bytes: int
+    upload_count: int
+    deduplicated_uploads: int