SEPA 0.1.0

New features

• simulate_sepa_fake_wj(): generate a synthetic WJ-IV look-alike dataset.
• sepa_stats_all(): compute plane-fit rho and direction cosines.
• parallel_analysis_ipsatized(): permutation-based dimensionality selection for ipsatized data.
• boot_cis_all(): BCa (with percentile fallback) CIs for all bootstrap indices from a boot object.
• percentile_ci_mat(): percentile CIs from a matrix of bootstrap draws.
• draw_sepa_biplot(): base-R row-isometric SVD biplot with optional highlighting and PDF output.
• write_long_to_wide(), write_matrix_wide(): tidy data reshaping helpers.
• run_sepa(): end-to-end wrapper executing the full SEPA pipeline.
• fake_wj: bundled 5 127-case synthetic dataset.

Bug fixes / clean-up (relative to development script)

• Loop variable renamed pkg (was p, conflicted with ncol variable).
• Duplicate cos_index() header comments removed.
• get_domain_names() defined once only.
• sepa_stats_all() defined once only.
• out object built once; redundant calls removed.
• drop = TRUE → drop = FALSE in all multi-column Procrustes slices.
• planes character vector renamed plane_labels (was overwriting list).
• pid initialised unconditionally.
• <<- superassignment removed from save_pdf(); restructured cleanly.
• install.packages() removed from body.
• All global-variable references in bootstrap statistics replaced with explicit arguments.