MAINT: Avoid NumPy deprecations

[larsoner]

Avoid at least some warnings of the form:

  /home/larsoner/python/nilearn/nilearn/glm/regression.py:100: DeprecationWarning: `np.float` is a deprecated alias for the builtin `float`. To silence this warning, use `float` by itself. Doing this will not modify any behavior and is safe. If you specifically wanted the numpy scalar type, use `np.float64` here.
  Deprecated in NumPy 1.20; for more details and guidance: https://numpy.org/devdocs/release/1.20.0-notes.html#deprecations

[codecov]

Codecov Report

Merging #2695 (a0aafd7) into master (29f64ef) will increase coverage by 0.02%.
The diff coverage is 100.00%.

@@            Coverage Diff             @@
##           master    #2695      +/-   ##
==========================================
+ Coverage   87.58%   87.61%   +0.02%     
==========================================
  Files         100      100              
  Lines       13164    13164              
  Branches     2555     2555              
==========================================
+ Hits        11530    11533       +3     
+ Misses       1026     1024       -2     
+ Partials      608      607       -1     

Impacted Files	Coverage Δ
nilearn/glm/first_level/first_level.py	74.38% <100.00%> (ø)
nilearn/glm/first_level/hemodynamic_models.py	95.28% <100.00%> (ø)
nilearn/glm/regression.py	75.69% <100.00%> (ø)
nilearn/reporting/_get_clusters_table.py	100.00% <0.00%> (+3.19%)	⬆️

---

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 29f64ef...ba4cab2. Read the comment docs.

[NicolasGensollen]

LGTM, thanks!

[bthirion]

Thx, this is great ! But I see many more of those in the codebase...

[larsoner]

The only ambiguity I see is whether to convert np.int to int (32-bit on Windows) or np.int64. If you're okay with me doing np.int64 everywhere I can probably make the rest of the substitutions

[bthirion]

I looked a bit superficially, and my first impression is that int would do the job in most places.

[larsoner]

Looks like these got implemented elsewhere by now