Restore Localizer fetchers interface for backwards compatibility

[kchawla-pi]

Fixes #2181

Making fetch_localizer_button_task & fetch_localizer_calculation_task backwards compatible.

• TODO: fetch_localizer_calculation_task()
• TOFIX: Example plot_dim_plotting.py is failing here & in pre PR 2097.

[codecov]

Codecov Report

Merging #2182 into master will decrease coverage by 0.15%.
The diff coverage is 100%.

@@            Coverage Diff             @@
##           master    #2182      +/-   ##
==========================================
- Coverage   92.34%   92.18%   -0.16%     
==========================================
  Files         149      149              
  Lines       18867    18868       +1     
  Branches     2302     2299       -3     
==========================================
- Hits        17423    17394      -29     
- Misses        924      952      +28     
- Partials      520      522       +2

Impacted Files	Coverage Δ
nilearn/datasets/tests/test_func.py	99.41% <100%> (ø)	⬆️
nilearn/datasets/func.py	81.42% <100%> (+0.42%)	⬆️
nilearn/conftest.py	47.36% <0%> (-52.64%)	⬇️
nilearn/plotting/__init__.py	78.26% <0%> (-17.4%)	⬇️
nilearn/input_data/nifti_masker.py	87.6% <0%> (-4.96%)	⬇️
nilearn/_utils/compat.py	43.33% <0%> (-3.34%)	⬇️
nilearn/__init__.py	68.57% <0%> (-2.86%)	⬇️
nilearn/decoding/fista.py	87.65% <0%> (-2.47%)	⬇️
nilearn/regions/parcellations.py	94.28% <0%> (-1.91%)	⬇️
nilearn/image/tests/test_image.py	98.36% <0%> (-1.1%)	⬇️
... and 2 more

---

Continue to review full report at Codecov.

Legend - Click here to learn more
Δ = absolute <relative> (impact), ø = not affected, ? = missing data
Powered by Codecov. Last update bc6a019...7a9605c. Read the comment docs.

[kchawla-pi]

@jeromedockes I didn't spot any obvious changes in fetch_localizer_calculation_task. Would you point me in what's different?

[jeromedockes]

@jeromedockes I didn't spot any obvious changes in `fetch_localizer_calculation_task`. Would you point me in what's different?

maybe I got confused and only the button task changed. both functions should have the same interface: no `n_subjects` parameter, and return a single subject (not a list of subjects).

[kchawla-pi]

I am comparing fetch_localizer_calculation_task from e7b118b7 and they both have n_subjects and list as return.

[jeromedockes]

I am comparing `fetch_localizer_calculation_task` from `e7b118b7` and they both have n_subjects and list as return.

ok, it's annoying that these two functions do not have an identical interface but it was already like that then. the important thing is that they stay the same as in the previous release.

[kchawla-pi]

Yeah I was surprised to see that as well.

[jeromedockes]

with the stable release the returned dicts have:

calculation: cmaps: <class 'list'>, ext_vars: <class 'numpy.recarray'>, description: <class 'bytes'>
button: description: <class 'bytes'>, tmap: <class 'str'>, anat: <class 'str'>

[jeromedockes]

and the parameters:

fetch_localizer_button_task(data_dir=None, url=None, verbose=1)
fetch_localizer_calculation_task(n_subjects=1, data_dir=None, url=None, verbose=1)

[jeromedockes]

maybe they should both have:

  * cmaps: list of 1 image (useless, here for backwards compatibility with stable fetch_localizer_calculation_task)
  * cmap: nifti image
  * tmap: nifti image
  * anat: nifti image

this way it is backward compatible but both functions return the same things.

[kchawla-pi]

maybe they should both have:

That is a different PR however. I will open a new PR for that.
Anything else here for the backward compatibility?