Bump LLVM to llvm/llvm-project@5fcf907 (llvm#2810)

[FxImporter] make FxImporter to fit python<=3.9 (llvm#2802)

As that torch with py3.9 is also used widely.

[Stablehlo]fix CumsumInputDtypeInt32Module_basic on stablehlo backend. (

llvm#2797)

Code used for testing.For the location of CumsumInputDtypeInt32Module in
the repo you can see
[here](https://github.com/llvm/torch-mlir/blob/311b6b0286bfa016346bc7fd8b441bbd50216060/projects/pt1/python/torch_mlir_e2e_test/test_suite/basic.py#L4148).
```python
import torch
import torch_mlir

class CumsumInputDtypeInt32Module(torch.nn.Module):

    def __init__(self):
        super().__init__()

    def forward(self, val):
        return torch.ops.aten.cumsum(val, 1)
module = torch_mlir.compile(CumsumInputDtypeInt32Module(), [torch.randn(2, 7, 4).to(torch.int32)], output_type="stablehlo")
print(module.operation.get_asm())
```
After fixing the bugs.
```
module attributes {torch.debug_module_name = "CumsumInputDtypeInt32Module"} {
  func.func @forward(%arg0: tensor<2x7x4xi32>) -> tensor<2x7x4xi64> {
    %0 = stablehlo.constant dense<0> : tensor<i64>
    %1 = stablehlo.convert %arg0 : (tensor<2x7x4xi32>) -> tensor<2x7x4xi64>
    %2 = "stablehlo.reduce_window"(%1, %0) ({
    ^bb0(%arg1: tensor<i64>, %arg2: tensor<i64>):
      %3 = stablehlo.add %arg1, %arg2 : tensor<i64>
      stablehlo.return %3 : tensor<i64>
    }) {padding = dense<[[0, 0], [6, 0], [0, 0]]> : tensor<3x2xi64>, window_dilations = dense<1> : tensor<3xi64>, window_dimensions = dense<[1, 7, 1]> : tensor<3xi64>, window_strides = dense<1> : tensor<3xi64>} : (tensor<2x7x4xi64>, tensor<i64>) -> tensor<2x7x4xi64>
    return %2 : tensor<2x7x4xi64>
  }
}
```

CI: Fix Roll PyTorch CI failure at determining commit hash (llvm#2796)

Signed-Off By: Vivek Khandelwal <vivekkhandelwal1424@gmail.com>

Fix onnx importer to treat Constant values as static (llvm#2780)

Fixes  llvm#2764

In the case of OPT, there are ConstantOfShape ops whose input shape is
not static (that is, an initializer), but rather comes from a Constant
op. The importer can't handle such non-static input shapes.

The fix here is to create initializers for a subset of Constant ops
(ones with "value" attributes), so that their outputs can be used
statically. Additionally, there was no case for creating a splat of
int64, so I added that as well.

---------

Co-authored-by: Dave Liddell <dliddell@xilinx.com>

g++ build fix (llvm#2778)

Introduced in 704cfda of @wu-s-john 

g++ compiler error: 

Pooling.cpp:177:13: error: explicit specialization in non-namespace
scope ‘class

Design looks good, g++ is just freaking out for no good reason.
Un-nesting the template classes fixes the error.

We don't have g++ CI. This hopefully happens infrequently enough that we
can just fix manually. My service to those folks who really like
building with g++... :)

g++ build fix (llvm#2778)

Introduced in 704cfda of @wu-s-john 

g++ compiler error: 

Pooling.cpp:177:13: error: explicit specialization in non-namespace
scope ‘class

Design looks good, g++ is just freaking out for no good reason.
Un-nesting the template classes fixes the error.

We don't have g++ CI. This hopefully happens infrequently enough that we
can just fix manually. My service to those folks who really like
building with g++... :)

g++ build fix (llvm#2778)

Introduced in 704cfda of @wu-s-john 

g++ compiler error: 

Pooling.cpp:177:13: error: explicit specialization in non-namespace
scope ‘class

Design looks good, g++ is just freaking out for no good reason.
Un-nesting the template classes fixes the error.

We don't have g++ CI. This hopefully happens infrequently enough that we
can just fix manually. My service to those folks who really like
building with g++... :)

Initial commit of NonZero op (llvm#2766)

[ONNX][MLIR] add Identity op support (llvm#2754)