TP/quantization/weight loading refactor part 1 - Simplify parallel linear logic (#1181)

tests/engine/test_detokenize.py

@@ -5,6 +5,7 @@ from transformers import AutoTokenizer
 from vllm.transformers_utils.tokenizer import detokenize_incrementally
 
 TRUTH = [
+    # pylint: disable=line-too-long
     "Hello here, this is a simple test",
     "vLLM is a high-throughput and memory-efficient inference and serving engine for LLMs. It is designed to be used in production environments, where inference and serving",
     "我很感谢你的热情"