[FIX] Logistic regression: fix penalty argument for no regularization

[janezd]

Issue

Fixes #6885.

As of 1.3.2, Scikit used to accept 'none' as penalty. In 1.5.1 it expects None.

I unsuccessfully tried to find where, when and why it changed, but it doesn't really matter, I guess.

Description of changes

Replace 'none' with None.

Includes

• Code changes
• Tests

[codecov]

Codecov Report

All modified and coverable lines are covered by tests ✅

Project coverage is 88.26%. Comparing base (6c742f7) to head (f9f4e59).
Report is 17 commits behind head on master.

Additional details and impacted files

@@            Coverage Diff             @@
##           master    #6886      +/-   ##
==========================================
- Coverage   88.26%   88.26%   -0.01%     
==========================================
  Files         326      326              
  Lines       71184    71184              
==========================================
- Hits        62834    62833       -1     
- Misses       8350     8351       +1