agriDesignR 0.2.1

Inference and post-hoc enhancements

Correct treatment and covariate role resolution: continuous covariates (e.g. plant_density, initial_height) are strictly isolated from post-hoc slicing and grouping.
Harden Compact Letter Display (CLD) assignment with robust support for overlapping significance groups (e.g. ab, bc) and multcompView alignment.
Add explicit block_as = c("auto", "fixed", "random") control in suggest_model() and model fitting routines.
Align plot_publication() and experiment_report() to share identical post-hoc families, adjustment methods, and estimands.
Improve nlme compatibility for fixed-block models and variance weighting structures.
Add regression test suites in tests/testthat/test-posthoc-contracts.R.

Add continuous integration checks with R CMD check --as-cran on Linux, macOS, and Windows, including release, previous-release, and development R.
Add automated test-coverage reporting through covr and Codecov.
Expand user-facing documentation with four package vignettes covering design planning, mixed models, diagnostics and remedies, and the Shiny app.
Add contributor guidance, issue templates, and CRAN release notes.

Add a pkgdown site configuration with grouped reference documentation.
Replace the static test-count badge in the README with live CI and coverage badges.

Existing calls and S3 classes from version 0.1.0 remain compatible.
Add explicit block_as control for fixed versus random blocking and publication-plot post-hoc controls.
Correct treatment/covariate role propagation and compact-letter displays, including overlapping groups and simple-effect comparisons.

Initial GitHub release with experimental design planning, randomized layouts, model fitting, diagnostics, post-hoc comparisons, remedies, publication plots, reports, and a Shiny interface.