[Misc] Use envs module to get VLLM_DISABLED_KERNELS (#35776)

Signed-off-by: Martin Hickey <martin.hickey@ie.ibm.com>

vllm/model_executor/kernels/linear/__init__.py

@@ -13,7 +13,6 @@ or kernel implementation, add it to this __init__.py to maintain
 import stability.
 """
 
-import os
 from typing import TypeVar
 
 import torch
@@ -154,8 +153,7 @@ _KernelConfigT = TypeVar("_KernelConfigT", bound=ScaledMMLinearLayerConfig)
 def is_supported_and_can_implement_kernel(
     kernel: type[_KernelT], config: _KernelConfigT, compute_capability: int | None
 ) -> tuple[bool, str]:
-    # TODO: Fetch `VLLM_DISABLED_KERNELS` from vllm.envs instead.
-    if kernel.__name__ in os.environ.get("VLLM_DISABLED_KERNELS", "").split(","):
+    if kernel.__name__ in envs.VLLM_DISABLED_KERNELS:
         return False, f" {kernel.__name__} is disabled by environment variable"
 
     if compute_capability is None: