[Kernel] Delegate construction of FusedMoEQuantConfig to FusedMoEMethodBase subclasses (#22537)

Signed-off-by: Bill Nell <bnell@redhat.com>

vllm/model_executor/layers/quantization/awq_marlin.py

@@ -9,8 +9,10 @@ from torch.nn import Parameter
 import vllm.model_executor.layers.fused_moe  # noqa
 from vllm import _custom_ops as ops
 from vllm.logger import init_logger
+from vllm.model_executor.layers.fused_moe.config import (FusedMoEConfig,
+                                                         FusedMoEQuantConfig)
 from vllm.model_executor.layers.fused_moe.layer import (
-    FusedMoE, FusedMoEConfig, FusedMoEMethodBase, FusedMoeWeightScaleSupported,
+    FusedMoE, FusedMoEMethodBase, FusedMoeWeightScaleSupported,
     UnquantizedFusedMoEMethod)
 from vllm.model_executor.layers.linear import (LinearBase, LinearMethodBase,
                                                UnquantizedLinearMethod,
@@ -483,6 +485,10 @@ class AWQMoEMethod(FusedMoEMethodBase):
         if hasattr(layer, "w2_bias") and layer.w2_bias is not None:
             layer.w2_bias.data = marlin_permute_bias(layer.w2_bias)
 
+    def get_fused_moe_quant_config(
+            self, layer: torch.nn.Module) -> Optional[FusedMoEQuantConfig]:
+        return None
+
     def apply(
         self,
         layer: torch.nn.Module,