BUG FIX project_to_curve(): Return error message when x or s contain insufficient rows.

BUG FIX unit tests: Switch from svg() to pdf() as support for svg() might be optional.

Fix warning in stats::approx() due to changes made in R 3.6.

Defuncted principal.curve() and get.lam().

Removed extra dependencies in princurve by removing vignettes; fixes #28.

Fully deprecated principal.curve() and get.lam().

DOCUMENTATION: Use the magick package to generate animated GIFs in the vignette, instead of the animation package, because animation uses ffmpeg which is not installed on all CRAN systems.

DEPRECATION: Added deprecation which will be triggered starting from 2018-08-01 upon calling principal.curve() or get.lam().

DOCUMENTATION: Added vignettes on the algorithm behind princurve and on benchmarking results between princurve 1.1 and 2.1.

BUG FIX principal_curve(): Don't apply rownames to curve as approx_points could be set to a different value other than nrow(x).

TESTING: Skip comparison unit test between princurve 1.1 and 2.1 on CRAN.

MINOR CHANGE project_to_curve(): Attempt to fix rchk warnings by not using x(i, j) = v notation but instead x[j * x.nrow() + i] = v.

DOCUMENTATION: Fix in README documentation.

BUG FIX principal_curve(): avoid division by zero when the initial principal curve has already converged.

BUG FIX project_to_curve(): set dimension names of outputted s correctly.

DOCUMENTATION: Added cran-comments.md and revdep to repository.

MINOR CHANGE: Removed adjust_range(); use grDevices::extendrange() instead.

TESTING start_circle(): Added unit tests.

BUG FIX start_circle(): Make sure circle is centered and scaled correctly.

MINOR CHANGE: Move smoother functions from inside principal_curve() to a list smoother_functions.

TESTING smoother_functions: Added tests to ensure each of the smoother functions work correctly.

SPEED UP project_to_curve(): Significantly speed up this function by not allocation objects that don't need allocation, and pre-allocating objects that do.

SPEED UP principal_curve(): Added approx_points parameter. This allows approximation of the curve between smoothing and projection, to ensure principal_curve() scales well to higher numbers of samples.