fix: stop grouping HF/ModelScope models by repository

A repository is not a model identity: collection repos on Hugging Face
and ModelScope host many unrelated models, which were wrongly shown as
versions of each other.

- Hugging Face models no longer auto-group (the Hub exposes no
  site-native model id)
- ModelScope models group by the site's native published-model id
  (MuseInfo modelVersion.modelId), extracted during enrichment and
  persisted on the sidecar as source_model_id/source_version_id;
  unenriched models stay standalone instead of collapsing a whole repo
  into one group
- TensorArt grouping unchanged (its URL id is already model-level)
- Frontend group-key derivation mirrors the new backend semantics

py/services/model_scanner.py

@@ -399,10 +399,14 @@ class ModelScanner:
             'skip_metadata_refresh': bool(get_value('skip_metadata_refresh', False)),
             # External model source (Hugging Face / ModelScope / TensorArt).
             # `source_url` + `source_platform` are canonical; `hf_url` stays in
-            # sync as a legacy alias (normalised below).
+            # sync as a legacy alias (normalised below).  `source_model_id` /
+            # `source_version_id` are the site-native identity ids version
+            # grouping keys off (ModelScope; empty elsewhere).
             'source_platform': get_value('source_platform', '') or '',
             'source_url': get_value('source_url', '') or '',
             'hf_url': get_value('hf_url', '') or '',
+            'source_model_id': get_value('source_model_id', '') or '',
+            'source_version_id': get_value('source_version_id', '') or '',
         }
         normalize_metadata_source(entry)