Description

Allow users to pass random_state in MapieQuantileRegressor constructor.

When calling fit(), the following logic is applied:

• If a random_state was set for this instance, and no random_state is passed in the function call, use the random state set in the constructor
• If no random_state is provided neither in fit nor constructor, no random_state is used
• If a random_state is provided in both fit, and constructor, the random_state from fit is used

My reasoning is that the random_state provided in fit will overwrite the one previously provided. This doesn't feel super intuitive to me (maybe the random state should not be passable now during fit to not have random state mismatches?), but I think it is better compared to the alternative

Fixes #405

Type of change

• Bug fix (non-breaking change which fixes an issue)
• New feature (non-breaking change which adds functionality)

How Has This Been Tested?

• Added a small unit test to make sure that the random state that is set when instancing results in the same output as a random state set during fit(), mimicking the desired behavior

Checklist

• I have read the contributing guidelines
• I have updated the HISTORY.rst and AUTHORS.rst files
• Linting passes successfully : make lint
• Typing passes successfully : make type-check
• Unit tests pass successfully : make tests
• Coverage is 100% : make coverage
• Documentation builds successfully : make doc