AHTauDesign 1.0

Major update

• Expanded AHTauDesign from truncation-time assessment to prospective trial planning for average hazard estimands.
• Added null calibration and support-rule freezing.
• Added simulate_design() as a public interface for generating simulation results.
• Implemented Stage 1 support audits and conditional Stage 2 horizon selection.
• Added independent evaluation of selected planning alternatives.
• Added Part II reverse design with coarse search, fine localization, and independent validation.
• Added auditable planning reports.

API changes

• Replaced the earlier assess_tau(), optimize_tau(), and twostage_tau() workflow with the new planning interfaces.
• Removed legacy IR outputs from Stage 2.
• Standardized the signed power comparison as power_difference.
• Added input validation and expanded automated testing.

Important notes

• Information support and statistical power are evaluated separately.
• A supported clinical horizon is retained.
• Alternative horizons are considered only after Stage 1 support failure.
• Support-rule thresholds require setting-specific calibration or validation.