add LayerNorm, RMSNorm, GroupNorm

[chaoming0625]

This pull request includes updates to the brainstate/nn/_elementwise/_dropout.py file to improve the handling of dropout probability and mask generation during the fitting process.

Changes to dropout probability handling:

• Updated the assertion in the __init__ methods to allow a dropout probability of 1.0, changing the range from [0, 1) to [0, 1]. [1] [2]

Changes to mask generation:

• Modified the __call__ method to include a condition that checks if the dropout probability is less than 1.0 before generating the mask during the fitting phase.
• Updated the init_state method to include a similar condition, ensuring the mask is generated only if the dropout probability is less than 1.0 during the fitting phase.