Fix wrong torch usage in communication wrapper for Distributed Inference Benchmark (#505)

**Description**
This commit fixes wrong `torch.empty_like` usage and missing dtype and
device argument in communication wrappers.

superbench/benchmarks/micro_benchmarks/dist_inference.py

@@ -121,7 +121,7 @@ def __all_gather_wrapper(self, x):
         Return:
             Tensor after all-gather.
         """
-        output = torch.empty_like([x.shape[0] * self.num_ranks] + list(x.shape[1:]))
+        output = torch.empty([x.shape[0] * self.num_ranks] + list(x.shape[1:]), dtype=x.dtype, device=x.device)
         dist.all_gather_into_tensor(output, x)
         return output