docs(attention): document None default behavior in MultiHeadAttention layers

[Ultron09]

What does this PR do?

Fixes #4182

In MultiHeadAttention and MultiHeadDotProductAttention (in flax.linen.attention) as well as MultiHeadAttention (in flax.nnx.nn.attention), several optional arguments default to None, but their docstrings did not document the exact fallback evaluation behavior when None is provided or left as default.

This PR clarifies the None default behavior for:

• qkv_features: defaults to inputs_q.shape[-1] (in Linen) or in_features (in NNX).
• out_features: defaults to inputs_q.shape[-1] (in Linen) or in_features (in NNX).
• deterministic: clarifies behavior when None vs dropout_rate > 0.0.
• precision: clarifies that None uses the default JAX numerical precision.

Checklist

• This PR fixes a minor issue (e.g.: typo or small bug) or improves the docs (you can dismiss the other checks if that's the case).
• This change is discussed in a Github issue/discussion.
• The documentation and docstrings adhere to the documentation guidelines.
• This change includes necessary high-coverage tests. (No quality testing = no merge!)

cc @vfdev-5 @cgarciae for review.