[Feat] Unquantized Linear to nz and control all nz-cast (#3356)

### What this PR does / why we need it?
Currently, when executing to the Linear layer of models in vLLM-Ascend,
the weights format is ND in unquantized case and skipped ascend case.
This PR supplements the execution logic for Linear layer. We use a new
global variable: VLLM_ASCEND_ENABLE_NZ. When VLLM_ASCEND_ENABLE_NZ=1 and
CANN version is 8.3, the weights of the Linear layer will be converted
to FRACTAL_NZ, in both unquantized case and skipped ascend case. We also
use VLLM_ASCEND_ENABLE_NZ to control the existing NZ conversion, such as
w8a8-quantized case.

### Does this PR introduce _any_ user-facing change?
Add a new global variable VLLM_ASCEND_ENABLE_NZ. If you want to use NZ
format, you should set VLLM_ASCEND_ENABLE_NZ=1.

### How was this patch tested?

- vLLM version: v0.11.0rc3
- vLLM main: https://github.com/vllm-project/vllm/commit/v0.11.0

Signed-off-by: anon189Ty <Stari_Falcon@outlook.com>

tests/ut/attention/test_mla_v1.py

@@ -376,7 +376,8 @@ class TestAscendMLAImpl(TestBase):
         self.assertEqual(q_pe.shape[1], self.impl.num_heads)
         self.assertEqual(q_pe.shape[2], self.impl.qk_rope_head_dim)
 
-    def test_process_weights_after_loading(self):
+    @patch('torch_npu.npu_format_cast')
+    def test_process_weights_after_loading(self, mock_format_cast):
         layer = MagicMock(spec=LinearBase)
         layer.input_size_per_partition = 10
         quant_method = MagicMock()
@@ -389,6 +390,7 @@ class TestAscendMLAImpl(TestBase):
         layer.weight = torch.randn(shape_0, shape_1)
         self.impl.kv_b_proj = layer
         apply.return_value = layer.weight.T
+        mock_format_cast.return_value = layer.weight
         self.impl.process_weights_after_loading(torch.bfloat16)
 
         self.assertEqual(self.impl.W_UK_T.shape[0], self.impl.num_heads)