[circle-mlir] Apply ONNX-MLIR upgrade

[seanshpark]

This will add changes for ONNX-MLIR and llvm-project upgrade.
Most of the changes are from llvm-project changes.

[seanshpark]

• for [circle-mlir] Upgrade ONNX-MLIR and other externals #15148
• from DRAFT circle-mlir version up externals #15153

[shs-park]

Now I've just triggered a workflow building new docker image.
Let's wait and retry the PR check then.

• https://github.com/Samsung/ONE/actions/runs/14482894843

[shs-park]

I've just updated the Docker image manually.
Let's try re-running the PR check.

[shs-park]

@seanshpark,
After docker image updated, it gets onnx2circle_value_test error.

        Start 104: onnx2circle_value_test_Add_F32_R4
104/521 Test #104: onnx2circle_value_test_Add_F32_R4 .....................................***Failed    0.10 sec
======================================================================
Run /__w/ONE/ONE/circle-mlir/build/Release/circle-mlir/tools-test/onnx2circle-value-test/exec_onnx.py /__w/ONE/ONE/circle-mlir/build/Release/circle-mlir/tools-test/onnx2circle-value-test/Add_F32_R4.onnx
Traceback (most recent call last):
  File "/__w/ONE/ONE/circle-mlir/build/Release/circle-mlir/tools-test/onnx2circle-value-test/exec_onnx.py", line 6, in <module>
    import h5py
  File "/usr/local/lib/python3.10/dist-packages/h5py/__init__.py", line 25, in <module>
    from . import _errors
  File "h5py/_errors.pyx", line 1, in init h5py._errors
ValueError: numpy.dtype size changed, may indicate binary incompatibility. Expected 96 from C header, got 88 from PyObject

        Start 105: onnx2circle_value_test_Add_F32_R4_C1
105/521 Test #105: onnx2circle_value_test_Add_F32_R4_C1 ..................................***Failed    0.10 sec
======================================================================

[shs-park]

Re-running PR checks with the docker image of #15173

[shs-park]

There are many warnings 😭

...

/__w/ONE/ONE/circle-mlir/circle-mlir/lib/pass/src/ops/HardSigmoidOp.h:61:45: warning: 'bool mlir::Type::isa() const [with Tys = {mlir::NoneType}]' is deprecated: Use mlir::isa<U>() instead [-Wdeprecated-declarations]
   61 |     if (alpha2.getType().isa<mlir::NoneType>())
      |         ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~^~
In file included from /workdir/include/mlir/IR/AttributeSupport.h:18,
                 from /workdir/include/mlir/IR/Attributes.h:12,
                 from /workdir/include/mlir/Bytecode/BytecodeImplementation.h:17,
                 from /workdir/include/mlir/Bytecode/BytecodeOpInterface.h:17,
                 from /workdir/include/mlir/IR/BuiltinOps.h:16,
                 from /workdir/include/mlir/Dialect/Func/Transforms/Passes.h:17,
                 from /__w/ONE/ONE/circle-mlir/circle-mlir/lib/pass/src/ConvertONNXToCirclePass.h:20,
                 from /__w/ONE/ONE/circle-mlir/circle-mlir/lib/pass/src/ConvertONNXToCirclePass.cpp:18:
/workdir/include/mlir/IR/Types.h:331:6: note: declared here

...

[shs-park]

Anyway all checks have passed
=)

[shs-park]

LGTM!
=)

[seanshpark]

There are many warnings 😭

Yes, warnings are from generated codes from *.td,
and the generator is circle-mlir/lib/tools/converter-gen/converter_gen.cc, from TF :)

[seanshpark]

@shs-park thanks! :)