Add test onnx.Cast bf16 to f16

[AmosLewis]

To test llvm/torch-mlir#3963

construct_inputs.log
Failed test at stage construct_inputs with exception:
Numpy doesn't support bfloat16. Please consider modifying the boundary types.

But if it pass manually run

iree-run-module --device=local-task --module=compiled_model.vmfb --function=main --input="1xbf16=1.0"
EXEC @main
result[0]: hal.buffer_view
1xf16=1

module {
  func.func @main(%arg0: !torch.vtensor<[1],bf16>) -> !torch.vtensor<[1],f16> attributes {torch.onnx_meta.ir_version = 10 : si64, torch.onnx_meta.opset_version = 21 : si64, torch.onnx_meta.producer_name = "", torch.onnx_meta.producer_version = ""} {
    %none = torch.constant.none
    %0 = torch.operator "onnx.Cast"(%arg0) {torch.onnx.to = 10 : si64} : (!torch.vtensor<[1],bf16>) -> !torch.vtensor<[1],f16> 
    return %0 : !torch.vtensor<[1],f16>
  }
}

#map = affine_map<(d0) -> (0)>
#map1 = affine_map<(d0) -> (d0)>
module {
  func.func @main(%arg0: tensor<1xbf16>) -> tensor<1xf16> {
    %0 = tensor.empty() : tensor<1xf16>
    %1 = linalg.generic {indexing_maps = [#map, #map1], iterator_types = ["parallel"]} ins(%arg0 : tensor<1xbf16>) outs(%0 : tensor<1xf16>) {
    ^bb0(%in: bf16, %out: f16):
      %2 = arith.extf %in : bf16 to f32
      %3 = arith.truncf %2 : f32 to f16
      linalg.yield %3 : f16
    } -> tensor<1xf16>
    return %1 : tensor<1xf16>
  }
}