Some failed ops when legalizing from torch to stablehlo

Nobody has claimed this yet.

Hi,
I'm experimenting with conversions of different pre-trained models to all 3 output dialects, and met 6 failures when converting to stablehlo. I'm showing the (simplified) examples below. Are this not implemented yet?
Note: some of these examples also fail when converting to tosa/linalg
Note2: examples with torch.constant depend on operations around the failing one: e.g. torch.constant.int 0 sometimes passes, sometimes fails.

1. error: failed to legalize operation 'torch.constant.int'
```
module attributes {torch.debug_module_name = "_lambda"} {
func.func @forward(%arg0: !torch.int) -> !torch.int {
%int0 = torch.constant.int 0
return %int0 : !torch.int
}
}
```

2.error: failed to legalize operation 'torch.constant.float'
```
module attributes {torch.debug_module_name = "_lambda"} {
func.func @forward(%arg0: !torch.float) -> !torch.float {
%float-Inf = torch.constant.float 0xFFF0000000000000
return %float-Inf : !torch.float
}
}
```

3.error: failed to legalize operation 'torch.aten.any'. (torch.constant.int 0 here seems ok!?)
```
module attributes {torch.debug_module_name = "_lambda"} {
func.func @forward(%arg0: !torch.vtensor<[2],si64>) -> !torch.vtensor<[],i1> {
%int0 = torch.constant.int 0
%0 = torch.aten.eq.Scalar %arg0, %int0 : !torch.vtensor<[2],si64>, !torch.int -> !torch.vtensor<[2],i1>
%1 = torch.aten.any %0 : !torch.vtensor<[2],i1> -> !torch.vtensor<[],i1>
return %1 : !torch.vtensor<[],i1>
}
}
```

4.error: failed to legalize operation 'torch.prim.ListConstruct'
```
module attributes {torch.debug_module_name = "_lambda"} {
func.func @forward(%arg0: !torch.vtensor<[1],si64>) -> !torch.list> {
%1 = torch.prim.ListConstruct %arg0 : (!torch.vtensor<[1],si64>) -> !torch.list>
return %1 : !torch.list>
}
}
```

5.error: failed to legalize operation 'torch.aten.detach'
```
module attributes {torch.debug_module_name = "_lambda"} {
func.func @forward(%arg0: !torch.vtensor<[1],si64>) -> !torch.vtensor<[1],si64> {
%0 = torch.aten.detach %arg0 : !torch.vtensor<[1],si64> -> !torch.vtensor<[1],si64>
return %0 : !torch.vtensor<[1],si64>
}
}
```

6.error: failed to legalize operation 'torch.aten.pow.Tensor_Tensor'
```
module attributes {torch.debug_module_name = "_lambda"} {
func.func @forward(%arg0: !torch.vtensor<[1,801,33],f32>, %arg1: !torch.vtensor<[],f32>) -> !torch.vtensor<[1,801,33],f32> {
%1 = torch.aten.pow.Tensor_Tensor %arg0, %arg1 : !torch.vtensor<[1,801,33],f32>, !torch.vtensor<[],f32> -> !torch.vtensor<[1,801,33],f32>
return %1 : !torch.vtensor<[1,801,33],f32>
}
}
```

I'm using standard command
torch-mlir-opt -pass-pipeline='builtin.module(torch-backend-to-stablehlo-backend-pipeline)'
All examples are reproduced on today's fresh torch-mlir.

Start by running torch-mlir-opt with the torch-backend-to-stablehlo-backend-pipeline on the six minimized MLIR examples in the issue. No source file or test is named; done would mean determining which reported operations are expected to legalize successfully and adding coverage for the confirmed failures.

Written by the indexing model from the issue text.