Add parity test for LayerNormalization

[tianleiwu]

Description:
Add test to verify precision of LayerNormalization.

Example output (result could be different since input is random):

Device	I/O Precision	LayerNorm Input Precision	MaxDiff	Comment
CPU	FP32	FP32	3.33e-06
CUDA	FP32	FP32	9.54e-07
CUDA	FP16	FP32	9.8e-04	Cast(to=FP32) -> LayerNorm(FP32) --> Cast(to=FP16)
CUDA	FP16	FP16	3.9e-03

Motivation and Context

• Why is this change required? What problem does it solve?
• If it fixes an open issue, please link to the issue here.