Run example llama2 model with fp16 (#1902)

Summary:
Pull Request resolved: https://github.com/pytorch/executorch/pull/1902

FYI - there are hardcoded `float` in rmsnorm which makes bunch of nodes in the graph as fp32.

```
aten_embedding_default: "f16[1, 3, 64]" = executorch_exir_dialects_edge__ops_aten_embedding_default(arg11_1, arg55_1);  arg11_1 = arg55_1 = None
aten_slice_copy_tensor: "f16[3, 4]"     = executorch_exir_dialects_edge__ops_aten_slice_copy_Tensor(arg48_1, 0, 0, 3);  arg48_1 = None
aten_slice_copy_tensor_1: "f16[3, 4]"   = executorch_exir_dialects_edge__ops_aten_slice_copy_Tensor(arg49_1, 0, 0, 3);  arg49_1 = None
aten__to_copy_default: "f32[1, 3, 64]"  = executorch_exir_dialects_edge__ops_aten__to_copy_default(aten_embedding_default, dtype = torch.float32)
(a lot of nodes in fp32 after this, and then we go back to fp16 and so on)
```

Copy op from - https://www.internalfb.com/code/fbsource/%5B7e45e7bcd969%5D/xplat/executorch/examples/models/llama2/model.py?lines=78

Reviewed By: larryliu0820

Differential Revision: D53596500

fbshipit-source-id: b6b3ebddfb9a25d1e52e9202d216e9ead9a6c62d