META-issue: deprecating features and bumping min requirements for 0.7.0

[rcap107]

I'm opening this issue to track a few deprecations and breaking changes that we need to consider.

Minimum requirements

• We need to bump scikit-learn requirements to 1.4 (currently it's 1.2)
• Python 3.9 stops being supported in October 2025, so we need to bump the minimum requirements to 3.10

Deprecations

• KEN embeddings Deprecate ken embeddings #1546
• DropIfTooManyNulls
• tabular_learner
• patch_display and unpatch_display
• specific_transformers parameter in TableVectorizer
• Values for .skb.apply(how=...) rename values for .skb.apply(how=...) #1628

These are most of the changes that I have found so far.

[rcap107]

Adding #1628

[rcap107]

Completed.
patch_display and unpatch_display were un-deprecated, and specific_transformers will be kept for the time being