Adds pipeline script for speed testing

[wpfl-dbt]

Context

We want to know to_hierarchical_clusters() can work at scale. This adds a pipeline script so we can easily test it.

• Download data to data/* from S3 or generate it withtest.fixtures.factories.generate_dummy_probabilities()
• Make sure the script references your data and run with uv run test/pipeline.py

My settings for small, medium, large:

hierarchical_cc2k = generate_dummy_probabilities(
    left_values=list(range(int(2e5))),
    right_values=list(range(int(2e5), int(4e5))),
    prob_range=(0.7, 1.0),
    num_components=2_000,
    total_rows=int(1e6)
)

hierarchical_cc20k = generate_dummy_probabilities(
    left_values=list(range(int(2e6))),
    right_values=list(range(int(2e6), int(4e6))),
    prob_range=(0.7, 1.0),
    num_components=20_000,
    total_rows=int(1e7)
)

hierarchical_cc200k = generate_dummy_probabilities(
    left_values=list(range(int(2e7))),
    right_values=list(range(int(2e7), int(4e7))),
    prob_range=(0.7, 1.0),
    num_components=200_000,
    total_rows=int(1e8)
)

Changes proposed in this pull request

• Adds test/pipeline.py to speed test stuff
• Adds rich for lovely progress bars

Guidance to review

• Read it
• Run it

Checklist:

• My code follows the style guidelines of this project
• New and existing unit tests pass locally with my changes

[leo-mazzone]

LGTM