[ONNX] Support embedding_renorm ONNX export
Composite using ONNX operators for same logic from here https://github.com/pytorch/pytorch/blob/0a07488ed2c47765e337e290bd138c0e6e459cbd/aten/src/ATen/native/Embedding.cpp#L153
Replaced #72560
Pull Request resolved: https://github.com/pytorch/pytorch/pull/72738
