torch.aten.convolution to linalg

[AmosLewis]

Find this issue in nod-ai/SHARK-TestSuite#189
Failed op in u-net_brain_mri_vaiq_int8.default.onnx.torch.elide.mlir

Failed op signature
%221 = torch.aten.convolution %212, %216, %220, %80, %59, %58, %true, %59, %int1 : !torch.vtensor<[32,512,16,16],f32>, !torch.vtensor<[512,256,2,2],f32>, !torch.vtensor<[256],f32>, !torch.list<int>, !torch.list<int>, !torch.list<int>, !torch.bool, !torch.list<int>, !torch.int -> !torch.vtensor<[32,256,32,32],f32>

To repeat the error

(mlir_venv) ➜  u-net_brain_mri_vaiq_int8 git:(reduceprod) ✗ torch-mlir-opt --torch-backend-to-linalg-on-tensors-backend-pipeline u-net_brain_mri_vaiq_int8.default.onnx.torch.mlir 
u-net_brain_mri_vaiq_int8.default.onnx.torch.mlir:238:12: error: failed to legalize operation 'torch.aten.convolution' that was explicitly marked illegal
    %221 = torch.aten.convolution %212, %216, %220, %80, %59, %58, %true, %59, %int1 : !torch.vtensor<[32,512,16,16],f32>, !torch.vtensor<[512,256,2,2],f32>, !torch.vtensor<[256],f32>, !torch.list<int>, !torch.list<int>, !torch.list<int>, !torch.bool, !torch.list<int>, !torch.int -> !torch.vtensor<[32,256,32,32],f32>

[zjgarvey]

I think I'm going to pick this up if no one else wants to dig into the quantization support issues.

If someone wants to take this off of my hands, let me know. It might take a day or two for me to start.

[zjgarvey]

Fails due to transposed convolutions not-yet implementing non-fp type transposes. I'll dig into this and see if I can make it work.

[zjgarvey]

torch-mlir PR #3240