Nested parallel loop is not supported by threadpool, try fuse them instead

Hi All,

I have a depthwise convolution in my model, and I enabled try_spatial_pack_depthwise = True. As soon as I enable try_spatial_pack_depthwise, I am getting following error when I do auto-tuning.

Any ideas?

!parallel_env_.in_parallel_loop: Nested parallel loop is not supported by threadpool, try fuse them instead