[oncall] fix flaky test (test_quant_dequant_not_fold) (#43)

Summary:
Pull Request resolved: https://github.com/pytorch/fx2trt/pull/43

Test `deeplearning.trt.fx2trt_oss.test.quant.test_quant_trt.TestQuantizeFxTRTOps` always fails due to incorrect sample inputs type. Fix it here.
See the task for more details.

Reviewed By: scottxu0730

Differential Revision: D35275923

fbshipit-source-id: 6695416a1942aa38c93c0a481f315479d7df9b66

Author: Cheng Cai

test/quant/test_quant_trt.py

@@ -752,7 +752,7 @@ def forward(self, x):
                 return self.relu(self.linear(x))
 
         model = LinearModule().eval()
-        inputs = torch.rand(8, 5)
+        inputs = [torch.rand(8, 5)]
 
         prepared = prepare_fx(model, {"": self.qconfig}, backend_config_dict=self.trt_backend_config_dict)
         quantized = convert_fx(