Description

There is a case

code
  x1 = torch.to(torch.ge(x0, 0.5), torch.device("cuda:0"), 3, False, False)
  _136 = torch.sum(torch.mul(x1, CONSTANTS.c0), [1], False)

graph
  %1284 : Tensor = prim::Constant[value=<Tensor>]()
  %1275 : Tensor = aten::ge(%x0.1, %1274)
  %x1.1 : Tensor = aten::to(%1275, %1315, %1278, %484, %484, %486)
  %1285 : Tensor = aten::mul(%x1.1, %1284)

The output of aten::ge is Bool type, and there are three way to support the above graph

1. implement aten::to, to convert datatype from Bool to Int
2. convert datatype in aten::mul (see Add support for bool in elementwise ops  #321), but I found the trtorch framework doesn't support specifying input data type, so I cann't test this method
3. convert the output type of Boolean Operation (that's what this PR does)

Type of change

• New feature (non-breaking change which adds functionality)
  Convert ouput type from bool to int

Checklist:

• My code follows the style guidelines of this project (You can use the linters)
• I have performed a self-review of my own code
• I have commented my code, particularly in hard-to-understand areas and hacks
• I have made corresponding changes to the documentation
• I have added tests to verify my fix or my feature
• New and existing unit tests pass locally with my changes