fix(typecheck): make the model_name stash env-independent

The `# type: ignore[assignment]` was correct locally and WRONG in CI. torch's
Module.__setattr__ is typed Tensor|Module, so `model.model_name = <str>` only
needs silencing when sentence-transformers/torch is actually installed; without
it the name is untyped and the ignore becomes an unused-ignore error. The gate
therefore failed in whichever environment you did not develop in — exactly the
local-vs-CI drift TOOL-PINS exists to catch, arriving by a different door.

setattr() is not checked against __setattr__ overloads, so it is clean either
way, and it is the honest spelling: the attribute is dynamic, declared by
neither SentenceTransformer nor torch.

mypy clean over 365 files locally (torch present); no ignore left to go unused
where it is absent.

Author: anthmFS

signalwire/signalwire/search/query_processor.py

@@ -291,10 +291,11 @@ def _get_cached_model(model_name: str | None = None) -> Any:
 
             logger.info(f"Loading sentence transformer model: {model_name}")
             model = SentenceTransformer(model_name)
-            # Store the model name for identification. torch's Module.__setattr__
-            # is typed for Tensor|Module, so stashing a plain str on the instance
-            # is a third-party-stub gap, not a real type error.
-            model.model_name = model_name  # type: ignore[assignment]
+            # Stash the name for identification. Dynamic: SentenceTransformer
+            # does not declare it, and torch's Module.__setattr__ is typed
+            # Tensor|Module, so a direct assignment only type-checks when torch
+            # is absent — setattr is correct whether or not it is installed.
+            setattr(model, "model_name", model_name)
             # Evict oldest entry if cache is full
             if len(_model_cache) >= _MAX_MODEL_CACHE_SIZE:
                 oldest_key = next(iter(_model_cache))

signalwire/signalwire/search/search_service.py

@@ -528,10 +528,10 @@ def _load_resources(self) -> None:
                             )
                             try:
                                 model = SentenceTransformer(model_name)
-                                # Store for cache comparison; torch's
-                                # Module.__setattr__ is typed Tensor|Module, so a
-                                # plain str is a stub gap, not a real error.
-                                model.model_name = model_name  # type: ignore[assignment]
+                                # Stash for cache comparison. Dynamic — see the
+                                # note in query_processor._load_model: a direct
+                                # assignment only type-checks when torch is absent.
+                                setattr(model, "model_name", model_name)
                                 self.models[model_name] = model
                             except Exception as e:
                                 logger.error(f"Failed to load model {model_name}: {e}")