feat: add drop parameter to OneHotEncoder (#913)

[BALOGUN-DAVID]

Closes #913

Description

This PR adds a drop parameter to the OneHotEncoder to allow users to control which dummy category is dropped when performing k-1 encoding. Previously, only the last category in the unique insertion order could be dropped via drop_last=True.

New drop strategies:

• 'last': Drops the last category in alphabetical order.
• 'first': Drops the first category in alphabetical order.
• 'most_frequent': Drops the most frequent category found during fit(). In the event of a tie, a UserWarning is raised and the first category alphabetically among the ties is dropped.

Backward Compatibility:
The drop_last parameter is fully preserved for backward compatibility. If both drop_last and drop are specified, drop takes precedence and a FutureWarning is emitted.

Checklist

• Added drop parameter and implementation logic
• Added FutureWarning when both drop and drop_last are set
• Updated docstrings for the new parameter
• Added comprehensive tests for all drop strategies and tie-breaking
• All tests and estimator checks pass

[BALOGUN-DAVID]

Hi @solegalli, it's me again! 👋

I've pushed the fixes and the PR is now ready for your review.

This adds the drop parameter to OneHotEncoder to allow dropping the 'first', 'last', or 'most_frequent' category, while preserving full backward compatibility for drop_last (as requested in #913). All tests and estimator checks are passing locally.

Let me know if you need any adjustments or if this is good to merge!