fix test failure on imports

Signed-off-by: qqma <qqma@amazon.com>

Author: qqma

vllm/v1/attention/backends/flash_attn.py

@@ -7,8 +7,8 @@
 import numpy as np
 import torch
 
-from vllm import envs
 from vllm import _custom_ops as ops
+from vllm import envs
 from vllm.attention.backends.abstract import (AttentionBackend, AttentionImpl,
                                               AttentionMetadata, AttentionType,
                                               is_quantized_kv_cache)