feat: add use_representation_model option to hierarchical_topics() for parent node labeling

[pidefrem]

What does this PR do?

feat: add use_representation_model option to hierarchical_topics() for parent node labeling

hierarchical_topics() builds parent topic names by concatenating the top 5 c-TF-IDF keywords (e.g., "wear_safety_PPE_worker_work"), while leaf topics get rich labels from the representation model (e.g., "PPE Non-Compliance Incidents" via an LLM). This creates a jarring inconsistency in visualize_hierarchy().

Root cause: _extract_words_per_topic is called with calculate_aspects=False for parent nodes.

Changes:

Add use_representation_model: bool = False parameter to hierarchical_topics(). When enabled, a post-processing step runs the full representation pipeline on all parent topics in a single batch call — not inline during the merge loop (which would make N-1 separate LLM calls).

# Before: parent nodes show "wear_safety_PPE_worker_work"
hierarchy = topic_model.hierarchical_topics(docs)

# After: parent nodes get proper labels from the representation model
hierarchy = topic_model.hierarchical_topics(docs, use_representation_model=True)

Default is False — existing behavior unchanged. The batch approach keeps LLM cost bounded.

Fixes #2503

Note: Cleaner if #2497 (shared helpers refactoring) lands first, but works standalone.

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?