[IMP] fs_attachment_s3_migration: checksum-based deduplication

Before this change, each attachment was processed individually causing
duplicate S3 uploads for files with identical content. The migration
domain was incomplete (missing type, checksum, db_datas filters), and
writing to the datas field risked data loss if source files were
unreadable.

After this change, attachments are grouped by checksum within each
batch - uploading each unique file once and updating all related
attachment records to point to the same S3 path. The domain now
properly excludes URL attachments, DB-stored files, and already
migrated files. Record updates use _force_write_store_fname instead
of datas writes, eliminating data loss risk. Added configurable
force-DB rules respect and updated  error logging.

Task: 5048

Author: Aldeigja

fs_attachment_s3_migration/data/queue_job_channel_data.xml

@@ -9,7 +9,7 @@
     To increase parallelism, configure in queue.job.channel settings or
     via server environment configuration:
 
-    Check https://github.com/OCA/queue/tree/18.0/queue_job documentation for more details.
+    Check https://github.com/OCA/queue/tree/16.0/queue_job documentation for more details.
 
     -->
     <record id="queue_channel_s3_migration" model="queue.job.channel">

fs_attachment_s3_migration/models/fs_storage.py

@@ -19,6 +19,13 @@ class FsStorage(models.Model):
         help="queue_job channel to use for migration jobs.",
     )
 
+    migration_use_storage_force_db_rules = fields.Boolean(
+        string="Use Storage Force-DB Rules",
+        default=True,
+        help="If checked, respect force_db_for_default_attachment_rules during "
+        "migration. Small images and assets will be skipped.",
+    )
+
     def action_open_migration_wizard(self):
         """Open the S3 migration wizard for this storage."""
         self.ensure_one()