homogenisation of docstring for the declaration of parameters

[lionelkusch]

For the moment, there are multiple ways to declare parameters:

version 1:
name : type, optional (default=default_value)
description
version 2:
name : type, default=default_value
description

What is the best way?
Do we indicate if a parameter is optional or not?

[bthirion]

The second one is the right one. We're currently propagating it throughout Nilearn.

[lionelkusch]

Do you want to apply it now, i.e. in the same as the model PR or to create a PR after just this point?

[bthirion]

Do it in several PRs to keep them light. It's a good task when you're tired ;-)

[lionelkusch]

ok. Thank you for the advice.

[lionelkusch]

File to refactors:

• ada_svr
• clustered_inference
• cpi
• dcrt
• desparsified_lasso
• Dnn_learner
• ensemble_clustred_inference
• gaussian_knockoff
• knockoff
• loco
• permutation_importance
• permutation_test
• RandomForestModified
• empirical_threshold
• utils/...