refactor(smoothquant): address HDCharles review comments

- Move is_distributed() guard to _apply_smoothing() hotpath so
  _reduce_activation_scales() only handles the distributed case,
  improving readability for single-GPU readers
- Remove redundant unit tests (subsumed by 2n_calls test; empty
  scales test unnecessary per reviewer feedback)
- Remove test_smoothquant_ddp_script_runs_cleanly (too expensive for CI)
- Switch integration test model to nm-testing/tinysmokellama-3.2
  (CI-friendly tiny model per HDCharles suggestion)
- Switch DDP example model to Qwen/Qwen2-7B-Instruct (DDP more
  meaningful for larger models)
- Fix --nproc arg conflict with torchrun, rename to --num_gpus
- Add benchmark_smoothquant_ddp.py for reproducing speedup numbers

Distributed speedup on 4x V100 32GB (Qwen2-7B-Instruct, 512 samples):
  1 GPU:  94.1 min | 8.93 GB peak mem | 1.00x
  2 GPU:  58.7 min | 7.06 GB peak mem | 1.60x
  4 GPU:  28.7 min | 7.06 GB peak mem | 3.28x

Addresses review comments from HDCharles on PR vllm-project#2471

Signed-off-by: David Zheng <dqzheng1996@gmail.com>

Author: David Zheng

examples/quantization_w8a8_int8/smoothquant_ddp_example.py

@@ -30,7 +30,7 @@
 # ---------------------------------------------------------------------------
 # Config
 # ---------------------------------------------------------------------------
-MODEL_ID = "TinyLlama/TinyLlama-1.1B-Chat-v1.0"
+MODEL_ID = "Qwen/Qwen2-7B-Instruct"
 DATASET_ID = "HuggingFaceH4/ultrachat_200k"
 DATASET_SPLIT = "train_sft"
 NUM_CALIBRATION_SAMPLES = 512

src/llmcompressor/modifiers/transform/smoothquant/base.py

@@ -343,7 +343,8 @@ def _apply_smoothing(self, model: Module):
 
         This modifies the weights of the model in-place.
         """
-        self._reduce_activation_scales()
+        if is_distributed():
+            self._reduce_activation_scales()
 
         for mapping in self.resolved_mappings_:
             if mapping.smooth_name not in self.scales_: