feat: Add Table One generator with privacy protection

[clifford-clif]

Summary

This PR adds a standardized Table 1 (baseline characteristics) generator with built-in privacy protection for federated research.

Features

• TableOneGenerator class for generating Table 1 summaries
• PrivacyConfig with cell suppression (default min_cell_size=10) and optional count rounding
• Support for continuous, categorical, and binary variable types
• DEFAULT_CLIF_VARIABLES for common ICU characteristics
• CSV and JSON export with metadata
• Comprehensive test suite

Privacy Protection

• Cell counts below threshold are suppressed (displayed as <10)
• Optional count rounding to nearest 5 or 10
• Percentage suppression when underlying count is suppressed
• Metadata tracks privacy settings for audit

Usage

from clifpy.utils.table_one import TableOneGenerator, PrivacyConfig

privacy = PrivacyConfig(min_cell_size=10, round_counts_to=5)
generator = TableOneGenerator(cohort_df, privacy_config=privacy, site_name="UCMC")
table1 = generator.generate()
generator.to_csv("table_one_UCMC.csv")

Why This Matters

This enables sites to generate consistent, privacy-safe Table 1 outputs for federated research where only aggregates are shared.

---

First PR from Clifford - CLIF AI assistant 🏥