update documentation

[sfluegel05]

Standardizes to numpy docstring format.

Claude says regarding the changes:

Summary

Unified the chebi_utils/ package on NumPy-style docstrings (Parameters/Returns), documenting inputs and outputs for every public method.

extract_properties.py (main work)

• Module # header → module docstring.
• Added full docstrings (folding in the inline # comment blocks) to get_atom_properties, get_bond_properties, get_molecule_level_properties, get_rings, get_steroid_positions.
• Added Parameters/Returns to mol_to_fol_atoms and get_numerical_facts.

sample_filters.py (main work)

• Module # header → module docstring.
• Full docstring for get_closest_negatives; added Parameters to get_direct_neighbors.
• Moved the local import queue to module top.

obo_extractor.py — converted get_hierarchy_subgraph's prose docstring to NumPy-style.

sdf_extractor.py — folded the "valid but no atoms" data-quirk # note into extract_molecules' docstring as a Notes section (keeping the ChEBI issue link).

Per your choices: private _-helpers were left untouched (public API only), and in-body # explanations were folded into the docstrings. dataset_builder.py and downloader.py already conformed and were left as-is.

Verification: ruff check and ruff format --check pass; all 65 tests pass (the initial failures were just two uninstalled runtime deps in this environment, now installed); help() spot-checks render the new Parameters/Returns sections.

[Copilot]

Pull request overview

This PR standardizes public-facing documentation across chebi_utils by converting module headers and public function docstrings to NumPy-style (Parameters/Returns/Notes), and updates the README to reflect the current public API and key features (including the validation split key).

Changes:

• Convert/expand public docstrings in extract_properties, sample_filters, and obo_extractor to NumPy-style sections.
• Move an important SDF parsing caveat into extract_molecules’ docstring Notes.
• Update README.md with the splits["validation"] breaking change note and add usage examples for FOL facts and hierarchy-aware sampling.

Reviewed changes

Copilot reviewed 5 out of 5 changed files in this pull request and generated 3 comments.

Show a summary per file

File	Description
README.md	Documents the validation split key breaking change and adds examples for property extraction and hierarchy-aware sampling.
chebi_utils/sdf_extractor.py	Moves record-dropping caveats into a Notes section in the function docstring.
chebi_utils/sample_filters.py	Adds NumPy-style docstrings and moves queue import to module scope.
chebi_utils/obo_extractor.py	Rewrites get_hierarchy_subgraph docstring into NumPy-style Parameters/Returns.
chebi_utils/extract_properties.py	Adds NumPy-style docstrings for public APIs related to FOL fact extraction.