Add aten::pixel_shuffle.out portable variant (#351)

Summary:
Pull Request resolved: https://github.com/pytorch/executorch/pull/351

The executorch portable runtime was missing an implementation of `aten::pixel_shuffle.out`,
which is used by the PiCA decoder.

This is basically just a reshape operator, and I adapted the implementation from
aten's `PixelShuffleKernel.cpp` after not using some helper functions that don't exist.

I don't know much about how to make a more optimized implementation, or if there's a way
to use parallelism automatically like the normal aten kernels do.

Reviewed By: manuelcandales

Differential Revision: D49173297

fbshipit-source-id: 625de6ec519b40929d4db9aae9c61fc1c49da691