M_noAria

Results 3 issues of M_noAria

Add dtype support and unit test. The old PR #2345 is a wrong implementation.

This PR replaces the direct torch operator of `pytest.mark.parametrize` with `getattr` in `test_torch_ops` to support parallel testing via `pytest-xdist`. --- Executing all unit tests on a single device takes too...

Adapt PyTorch 2.8.0 and fix unit tests. ## Current failed test cases ``` coremltools/converters/mil/frontend/torch/test/test_torch_conversion_api.py::TestTorchao::test_weight_only_quantization coremltools/converters/mil/frontend/torch/test/test_torch_export_conversion_api.py::TestExecuTorchExamples::test_add coremltools/converters/mil/frontend/torch/test/test_torch_export_conversion_api.py::TestExecuTorchExamples::test_add_mul coremltools/converters/mil/frontend/torch/test/test_torch_export_conversion_api.py::TestExecuTorchExamples::test_linear coremltools/converters/mil/frontend/torch/test/test_torch_export_conversion_api.py::TestExecuTorchExamples::test_mul coremltools/converters/mil/frontend/torch/test/test_torch_export_conversion_api.py::TestExecuTorchExamples::test_softmax coremltools/converters/mil/frontend/torch/test/test_torch_ops.py::TestActivation::test_selu coremltools/converters/mil/frontend/torch/test/test_torch_ops.py::TestBatchNorm::test_batchnorm_dynamic_stress coremltools/converters/mil/frontend/torch/test/test_torch_ops.py::TestEinsum::test_ternary_einsum coremltools/converters/mil/frontend/torch/test/test_torch_ops.py::TestIndexPut::test_index_put_negative_indices_case_2 coremltools/converters/mil/frontend/torch/test/test_torch_ops.py::TestIndexPut::test_index_put_updates_bool coremltools/converters/mil/frontend/torch/test/test_torch_ops.py::TestUnfold::test_unfold coremltools/converters/mil/frontend/torch/test/test_torch_quantization_ops.py::TestPytorchQuantizedOps::test_unpack_int4packed_by_mm_with_eye_matrix coremltools/converters/mil/frontend/torch/test/test_torch_quantization_ops.py::TestPytorchQuantizedOps::test_weight_int4pack_mm coremltools/test/optimize/torch/quantization/test_coreml_quantizer.py::test_weight_module_act_fusion ``` ##...

feature request