fix: pop model_tokens so it is not forwarded to the model client

On the plain llm-config path, _create_llm read model_tokens but never
removed it from llm_params, so it was splatted into init_chat_model and
forwarded to the client. With ChatOpenAI (any OpenAI-compatible base_url)
this surfaced as: Completions.create() got an unexpected keyword argument
'model_tokens'. Pop the key after consuming it, matching the model_instance
path which already returns before the key can leak.

Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>

Author: dylanpfirn

scrapegraphai/graphs/abstract_graph.py

@@ -221,6 +221,9 @@ def _create_llm(self, llm_config: dict) -> object:
         else:
             self.model_token = llm_params["model_tokens"]
 
+        # Consumed by ScrapeGraphAI; must not be forwarded to the model client.
+        llm_params.pop("model_tokens", None)
+
         try:
             if llm_params["model_provider"] not in {
                 "oneapi",