[Inference Providers] Raise warning if provider's status is in error mode (#3141)

* update hint

* add warning

* Update src/huggingface_hub/inference/_providers/_common.py

Co-authored-by: Lucain <[email protected]>

---------

Co-authored-by: Lucain <[email protected]>

Author: hanouticelina

src/huggingface_hub/hf_api.py

@@ -716,7 +716,7 @@ class InferenceProviderMapping:
     provider: "PROVIDER_T"  # Provider name
     hf_model_id: str  # ID of the model on the Hugging Face Hub
     provider_id: str  # ID of the model on the provider's side
-    status: Literal["live", "staging"]
+    status: Literal["error", "live", "staging"]
     task: str
 
     adapter: Optional[str] = None

src/huggingface_hub/inference/_providers/_common.py

@@ -144,6 +144,11 @@ def _prepare_mapping_info(self, model: Optional[str]) -> InferenceProviderMappin
             logger.warning(
                 f"Model {model} is in staging mode for provider {self.provider}. Meant for test purposes only."
             )
+        if provider_mapping.status == "error":
+            logger.warning(
+                f"Our latest automated health check on model '{model}' for provider '{self.provider}' did not complete successfully.  "
+                "Inference call might fail."
+            )
         return provider_mapping
 
     def _prepare_headers(self, headers: Dict, api_key: str) -> Dict: