Changes in version 0.1.1                        

  - Allow the nugget variance to be fixed independently while other
    covariance parameters are estimated. nugget_variance = 0 now fits
    the no-nugget boundary model, and estimate_nugget = FALSE fixes any
    supplied non-negative nugget.
  - Propagate fixed-nugget parameter maps through dense Gaussian,
    Laplace, Monte Carlo maximum-likelihood, nearest-predecessor,
    prediction-uncertainty, simulation-recovery, likelihood and
    reporting workflows.
  - Report fixed covariance parameters without invented standard errors
    or likelihood degrees of freedom.
  - Correct the transformed-covariance ordering used by predictive
    parameter draws when grouped random slopes and temporal scale are
    both estimated.
  - Add native and WebAssembly regression coverage for no-nugget models.

                        Changes in version 0.1.0                        

  - First tagged course release.