Changes in version 0.1.1 (2026-09-02)                  

  - Removed the URL and BugReports fields from DESCRIPTION: they pointed
    at a repository that does not exist and returned 404 in CRAN's URL
    check.
  - Fixed a false INCONSISTENT verdict in the degrees-of-freedom
    additivity check of bk_check_anova(). The comparison used
    isTRUE(all.equal()), which reported a difference on attribute
    mismatch even when the two totals were numerically identical; it now
    compares attribute-stripped numeric values against a tolerance.

                        Changes in version 0.1.0                        

  - Initial release.
  - Four audit modules, all restricted to definition-free identities:
      - bk_check_variability() — GCV/PCV/heritability/genetic-advance
        identities (h2 = (GCV/PCV)^2; GAM = K * h2 * PCV; GAM = K *
        GCV^2 / PCV; GCV <= PCV).
      - bk_check_anova() — df and SS additivity, MS = SS/df, F = MS/MSe.
      - bk_check_precision() — CV%, SEm, CD from MSe, and CD = t * SEm *
        sqrt(2).
      - bk_check_corr() — symmetry, unit diagonal, |r| <= 1, and
        positive semi-definiteness with a rigorous Weyl rounding bound.
  - bk_audit() combines module results; print() and summary() methods.
  - All comparisons use rounding-interval arithmetic: a value is flagged
    only when no combination of values inside the reported rounding
    intervals can satisfy the identity.
  - k = "auto" reconciles genetic advance against the standard selection
    differential family (2.64, 2.06, 1.76, 1.40) and reports which
    member fits.
  - Worked examples via bk_example(), including two planted errors that
    demonstrate error localisation.