Fix foreign key constraints and improve snapshot tracking

- Add unified compression/encryption package in internal/blobgen
- Update DATAMODEL.md to reflect current schema implementation
- Refactor snapshot cleanup into well-named methods for clarity
- Add snapshot_id to uploads table to track new blobs per snapshot
- Fix blob count reporting for incremental backups
- Add DeleteOrphaned method to BlobChunkRepository
- Fix cleanup order to respect foreign key constraints
- Update tests to reflect schema changes

internal/database/models.go

@@ -37,11 +37,9 @@ type FileChunk struct {
 
 // Chunk represents a data chunk in the deduplication system.
 // Files are split into chunks which are content-addressed by their hash.
-// The ChunkHash is used for deduplication, while SHA256 provides
-// an additional verification hash.
+// The ChunkHash is the SHA256 hash of the chunk content, used for deduplication.
 type Chunk struct {
 	ChunkHash string
-	SHA256    string
 	Size      int64
 }