torch.aten.squeeze to linalg

[saienduri]

We are seeing following issues when lowering into torch.aten.squeeze in these models (mobilebert-uncased). Repro instructions can be found here::

mobilebert-uncased.default.pytorch.torch.mlir:4676:13: error: failed to legalize operation 'torch.aten.squeeze' that was explicitly marked illegal
    %3207 = torch.aten.squeeze %3206 : !torch.vtensor<[1,1,512],f32> -> !torch.vtensor<[1,512],f32>
            ^
mobilebert-uncased.default.pytorch.torch.mlir:4676:13: note: see current operation: %39331 = "torch.aten.squeeze"(%39330) : (!torch.vtensor<[1,1,512],f32>) -> !torch.vtensor<[1,512],f32>
iree-compile: /home/sai/iree/third_party/llvm-project/mlir/include/mlir/IR/UseDefLists.h:198: mlir::IRObjectWithUseList<mlir::OpOperand>::~IRObjectWithUseList() [OperandType = mlir::OpOperand]: Assertion `use_empty() && "Cannot destroy a value that still has uses!"' failed.

[aldesilv]

resolved, upstream fix