                     Changes in version 2026-07-31                      

  - all functions refactored for performance and increased robustness
  - refactor calls to getb() to consistently pass 3d bcdiff array
  - barray: new value (value="bc-diff") returns a 3d difference array
    with values -1,0,1
  - bcluster: can now be called with value="bc-diff"
  - bcluster.h: refactoring only
  - bcluster.n: can now be called with value="bc-diff"
  - bcluster.n: more robust validation if M provided; M more random if M
    unspecified
  - bcdiff: new convenience function calls barray with value="bc-diff"
  - cochranQ: now returns unrounded results
  - getb: breaking change to operate on a 3d bcdiff array (thanks to M.
    Meyners who suggested this approach would improve performance)
  - getb: breaking changes with new method signature removes oneI and
    oneM parameters
  - evaluateClusterQuality: refactoring only
  - homogeneity: low-impact breaking change; oneI and oneM parameters
    removed
  - mcnemarQ: now returns unrounded results
  - plift: breaking changes with new method signature
  - plift: always provides verbose output and returns unrounded results
  - plift: new parameter gives option to drop flatlining assessors
    (default: drop=FALSE)
  - toMatrix: fully deprecated; replaced by toTallMatrix()
  - toTallMatrix: new function replaces toMatrix()
  - toTallMatrix: 'subject' header row in toMatrix() renamed to
    'assessor'
  - toTallMatrix: low-impact breaking change from toMatrix removes oneI
    and oneM parameters
  - toWideMatrix: includes dimension names

                      Changes in version 2025-9-11                      

  - remove data() calls from examples since data are lazy loaded
  - getb: separate to simplify refactoring for performance later
  - getb: update to maintain array dimensions
  - update reference from preprint doi:10.48550/arXiv.2502.15945 to
    published manuscript doi:10.1016/j.foodqual.2025.105639
  - bcluster.n: inner functions promoted to top-level functions to
    enable future unit tests
  - bcluster.n: cluster memberships identified using ordered cluster
    numbers
  - inspect: legacy cluster memberships checked and updated to use
    ordered cluster numbers
  - inspect: (bug fix) Index values in inspect function output corrected
    (thanks to M. Meyners for reporting this issue)
  - inspect: documentation update to indicate the parameter G used in
    hierarchical clustering
  - inspect: plot for non-hierarchical clustering results ignores G and
    plots results given
  - inspect: plot title updated to describe plot type and groups plotted
  - evaluateClusterQuality: (bug fix) array reshaping issue resolved to
    allow RV calculations
