[refactor] refactor weight trans nz and transpose (#4878)

### What this PR does / why we need it?

Now `VLLM_ASCEND_ENABLE_NZ` will have three options:
0: disable nz;
1: only quant case enable nz;
2: enable nz as long as possible;

And `VLLM_ASCEND_ENABLE_NZ`=1 by default.

All cases are shown in the table below:

|  | W4A4 | W4A8 | W8A8 | fp16/bf16 | fp32 |
|---|---|---|---|---|---|
| trans nz | can't support nz | trans nz by default | trans nz by
default | trans nz when VLLM_ASCEND_ENABLE_NZ is 2 | can't support nz |
| transpose | only support not transpose case | only support transpose
case | only support transpose case | linear: only support not transpose
case<br>gmm: only support transpose case | same to fp16/bf16 |

Some exceptional cases:
1. MLAPO op need to do some additional processing on the weights,
including trans nz. If use MLAPO op, some weight will be transformed to
nz forcely;
2. MLA/SFA's weight `W_UV` will be used by op
`torch.ops._C_ascend.batch_matmul_transpose`, and this op can't support
nz currently;

### Does this PR introduce _any_ user-facing change?
Now fp16/bf16 weight will not trans nz by default.

### How was this patch tested?

- vLLM version: v0.12.0
- vLLM main:
ad32e3e19c

Signed-off-by: zzzzwwjj <1183291235@qq.com>

vllm_ascend/ops/fused_moe/fused_moe.py

@@ -19,7 +19,6 @@ from typing import Any, Callable, Optional
 
 import torch
 import torch.nn.functional as F
-import torch_npu
 from vllm.config import get_current_vllm_config
 from vllm.distributed import (get_dp_group, get_ep_group, get_tp_group,
                               tensor_model_parallel_all_reduce)
@@ -48,8 +47,8 @@ from vllm_ascend.quantization.w4a8_dynamic import \
     AscendW4A8DynamicFusedMoEMethod
 from vllm_ascend.quantization.w8a8_dynamic import \
     AscendW8A8DynamicFusedMoEMethod
-from vllm_ascend.utils import (ACL_FORMAT_FRACTAL_NZ, AscendDeviceType,
-                               enable_sp, get_ascend_device_type, is_enable_nz,
+from vllm_ascend.utils import (AscendDeviceType, enable_sp,
+                               get_ascend_device_type, maybe_trans_nz,
                                npu_stream_switch, shared_expert_dp_enabled,
                                shared_experts_calculation_stream)
 
@@ -73,12 +72,9 @@ class AscendUnquantizedFusedMoEMethod(UnquantizedFusedMoEMethod):
             1, 2).contiguous()
         layer.w2_weight = torch.nn.Parameter(w2_data, requires_grad=False)
 
-        if get_ascend_device_type() != AscendDeviceType._310P and is_enable_nz(
-        ):
-            layer.w13_weight.data = torch_npu.npu_format_cast(
-                layer.w13_weight.data, ACL_FORMAT_FRACTAL_NZ)
-            layer.w2_weight.data = torch_npu.npu_format_cast(
-                layer.w2_weight.data, ACL_FORMAT_FRACTAL_NZ)
+        if get_ascend_device_type() != AscendDeviceType._310P:
+            layer.w13_weight.data = maybe_trans_nz(layer.w13_weight.data)
+            layer.w2_weight.data = maybe_trans_nz(layer.w2_weight.data)
 
     def apply(self,
               layer: torch.nn.Module,