refactor: extract shared helper methods for document aggregation and topic naming

[pidefrem]

What does this PR do?

refactor: extract shared helper methods for document aggregation and topic naming

Several methods in _bertopic.py contain near-identical copy-paste patterns that have diverged slightly over time:

1. Document aggregation — groupby("Topic").agg({"Document": " ".join}) appears in 4 methods (_extract_topics, hierarchical_topics, update_topics, partial_fit). When a new column needs aggregating, every call site must be updated independently.
2. Feature names extraction — a 5-line sklearn version check (get_feature_names vs get_feature_names_out) is duplicated in hierarchical_topics and _c_tf_idf.
3. Topic name construction — "_".join([x[0] for x in words][:N]) appears in 4 places across 2 methods with inconsistent slicing ([:4] in topic_labels_, [:5] in hierarchical_topics).

Changes:

Extract three private helper methods:

• _aggregate_documents() — centralizes the groupby/agg pattern
• _get_feature_names() — centralizes the sklearn version check
• _topic_name_from_words() — centralizes topic name string construction with consistent slicing

All existing call sites updated to use the helpers. Zero behavior change — pure refactoring. All existing tests pass without modification.

Fixes #2497

Before submitting

• This PR fixes a typo or improves the docs (if yes, ignore all other checks!).
• Did you read the contributor guideline?
• Was this discussed/approved via a Github issue? Please add a link to it if that's the case.
• Did you make sure to update the documentation with your changes (if applicable)?
• Did you write any new necessary tests?