EnTraineR

_First major release._

This version consolidates the API, improves robustness for **FactoMineR** outputs, and harmonizes audience-tailored prompts across all functions.

Major Changes & New Features

Unified Architecture (“Hub & Spoke”)

All trainer functions now rely on a centralized core (trainer_core.R) for consistent **prompt building**, **audience profiling**, **output capture**, and **regex/quoting helpers**. Any update to the core instantly benefits all trainers.

New Trainers

• trainer_t_test() — Interpret Student’s t-tests (one-sample, two-sample, paired, Welch) with audience-aware guidance on **p-values vs confidence intervals**.
• trainer_cor_test() — Interpret correlation tests (**Pearson**, **Spearman**, **Kendall**) with clear distinction between **statistical significance** and **practical magnitude**.
• trainer_var_test() — Interpret F-tests for equality of variances.

Robust ANOVA & Linear Model Support

• trainer_AovSum() and trainer_LinearModel() use **centralized heuristics** (trainer_core_extract_tables_heuristic) to reliably extract **F-test** and **T-test** tables even when standard headers are missing (common with capture.output).
• **Strict, space-safe T-table filtering**: factor names with spaces or special characters are handled via core quoting (trainer_core_quote) and detection helpers.