DOC: Fix non-standard napoleon custom section mapping

[jhlegarreta]

Fix non-standard napoleon custom section mapping:

• Map Inputs and Outputs to params_style and returns_style, respectively, instead of mapping them to Parameters.
• Map Mandatory Inputs and Optional Inputs to params_style, instead of mapping them to Parameters.
• Add the numpydoc extension and set the numpy_doc_show_class_members and autodoc_typehints flags so that the Attributes docstrings are shown.
• Map the Attributes section to the Attributes string in order to show that value instead of Variables in the corresponding documentation section.

Documentation:
https://www.sphinx-doc.org/en/master/usage/extensions/napoleon.html#confval-napoleon_custom_sections https://stackoverflow.com/questions/72220924/sphinx-how-to-show-attributes-as-in-scipy

[welcome]

Thanks for opening this pull request! We have detected this is the first time for you to contribute to dMRIPrep. Please check out our contributing guidelines.
We invite you to list yourself as a dMRIPrep contributor, so if your name is not already mentioned, please modify the .zenodo.json file and insert your data last in the contributors list. Example:

{
   "name": "Contributor, New dMRIPrep",
   "affiliation": "Department of dMRI prep'ing, Open Science Made-Up University",
   "orcid": "<your id>",
   "type": "Researcher"
}

Of course, if you want to opt-out this time there is no problem at all with adding your name later. You will be always welcome to add it in the future whenever you feel it should be listed.

[jhlegarreta]

Comments:

• Related to DOC: Fix non-standard Input/Output docstring section management petprep#105 and DOC: Fix non-standard Input/Output docstring section management fmriprep#3505
• I am not sure if all those flags are necessary for the Attribute thing to work: I have taken them from the linked SO. The more elegant napoleon_use_ivar = True solution seems to display the attributes under a Variables section, which is not very intuitive.
  • Tried locally on nifreeze and it did not work either way.