Refactor: Centralize local model initialization in _BaseModel

[jigyasaba]

Description:

While reviewing the BaseClassifier and BaseRegressor implementations, I noticed the logic for inspecting random_state during local model initialization was duplicated.

I have refactored this into a helper method _initialize_local_model in the parent _BaseModel class. This adheres to the DRY principle and simplifies the _fit_local loop in both subclasses, making the codebase easier to maintain as we add new modules.

Validation:

Verified that random_state is correctly passed to models that support it.

Ran pytest gwlearn/tests/test_base.py and all 112 tests passed.

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 93.09%. Comparing base (34ab4fd) to head (8e34a5c).

Additional details and impacted files

@@            Coverage Diff             @@
##             main      #99      +/-   ##
==========================================
+ Coverage   92.96%   93.09%   +0.12%     
==========================================
  Files           6        6              
  Lines         825      825              
==========================================
+ Hits          767      768       +1     
+ Misses         58       57       -1     

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.

[martinfleis]

It is duplicated but explicit. Anyone reading the code immediately gets it without figuring out what does the private method do.

Any PR shall have a linked and approved issue. Closing as this one does not.

[jigyasaba]

Thank you for the clarification, I understand the preference for explicitness in the fit loops over the DRY refactor here. I also appreciate the reminder on the workflow—I will ensure all future PRs are linked to an approved issue first.