Remove useless env (#4858)

cleanup useless env. These envs are not used anymore

`VLLM_ASCEND_TRACE_RECOMPILES`,
`VLLM_ASCEND_KV_CACHE_MEGABYTES_FLOATING_TOLERANCE`,
`VLLM_ASCEND_MLA_PA`, `PHYSICAL_DEVICES`

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

Signed-off-by: wangxiyuan <wangxiyuan1007@gmail.com>

tests/e2e/multicard/test_offline_inference_distributed.py

@@ -138,7 +138,6 @@ def test_models_distributed_Qwen3_W4A8DYNAMIC_new_version(model):
 
 
 @pytest.mark.parametrize("model", DEEPSEEK_W4A8_MODELS)
-@patch.dict(os.environ, {"VLLM_ASCEND_MLA_PA": "1"})
 @patch.dict(os.environ, {"HCCL_BUFFSIZE": "1024"})
 def test_models_distributed_DeepSeek_W4A8DYNAMIC(model):
     prompts = [

tests/ut/kv_connector/test_mooncake_connector.py

@@ -1055,7 +1055,6 @@ class MockTransferEngine:
 
 class MockEnvsAscend:
     MOONCAKE_CONNECTOR_PROTOCOL = "mock_protocol"
-    PHYSICAL_DEVICES = "10,11"
 
 
 def mock_get_tensor_model_parallel_rank():

tests/ut/kv_connector/test_mooncake_layerwise_connector.py

@@ -893,12 +893,3 @@ class TestMooncakeLayerwiseConnectorWorker(unittest.TestCase):
         worker.register_kv_caches(mla_caches)
         self.assertTrue(worker.use_mla)
         self.assertEqual(len(worker.block_len), 2)
-
-    def test_device_id_selection_with_physical_devices(self):
-        worker = MooncakeLayerwiseConnectorWorker(self.vllm_config,
-                                                  self.engine_id)
-        self.assertIsNotNone(worker.engine)
-
-
-if __name__ == '__main__':
-    unittest.main()