Restore consistency of hook_normalized between LayerNorm and RMSNorm

[degenfabian]

Description

This PR intends to fix issue #747. hook_normalized is applied after the gain and bias weights are used in layer_norm.py, whereas in rms_norm.py it's before. This inconsistency was fixed by moving hook_normalized before the gain and bias weights in layer_norm.py. According to @neelnanda-io this will be a breaking change, which after @bryce13950 could be worked into release 3.0. I figured since there was already so much guidance as to how to go about the change in the issue itself and since the actual change was so small, I would just go ahead and do it. I would greatly appreciate some guidance about whether or not I should add tests for this, since I am new to mech Interp and this library and wouldn't know how to go about that.

Thanks for maintaining this library!

Fixes # (issue) #747

Type of change

• Breaking change (fix or feature that would cause existing functionality to not work as expected)

Checklist:

• I have commented my code, particularly in hard-to-understand areas
• I have made corresponding changes to the documentation
• My changes generate no new warnings
• I have added tests that prove my fix is effective or that my feature works
• New and existing unit tests pass locally with my changes
• I have not rewritten tests relating to key interfaces which would affect backward compatibility

[bryce13950]

Thanks for putting this together! This is going to sit open for a while until we start testing some changes on 3.0. When we are ready to do that, I will get it merged right away!