Convert `benchmarks` to `ruff format` (#18068)

Signed-off-by: Harry Mellor <19981378+hmellor@users.noreply.github.com>

benchmarks/kernels/deepgemm/benchmark_fp8_block_dense_gemm.py

@@ -11,7 +11,9 @@ from deep_gemm import calc_diff, ceil_div, get_col_major_tma_aligned_tensor
 # Import vLLM functions
 from vllm import _custom_ops as ops
 from vllm.model_executor.layers.quantization.utils.fp8_utils import (
-    per_token_group_quant_fp8, w8a8_block_fp8_matmul)
+    per_token_group_quant_fp8,
+    w8a8_block_fp8_matmul,
+)
 from vllm.triton_utils import triton