v3.0.0

• Fitting to specific polynomials instead of the full range is now allowed using an input vector of (m, n) tuples containing the subset;
• Implemented a wavefront error matrix input method;
• The Model methods are now dispatched using parametric type selectors;
• The scale keyword option has been renamed to finesse;
• The new stack function is now used to efficiently construct the phase error fitting matrix;
• Slightly optimized the radial polynomial coefficient algorithm;
• Fixed conversion and indexing errors for wavefront error fitting under n_max = 0 (mean piston output);
• Log transforms are now constructed more efficiently using broadcasted assignment;
• Adjusted the allowed input types for the wavefront error modelling;
• The window and plot title keyword arguments are now optional;
• Added a phase mapping function which converts a uniformly sampled set of data to coordinate vectors and a phase matrix;
• Computation of the RMS wavefront error has been improved by using the inner product;
• Validation of the pupil scaling factor is now carried out independent of translation.

Full Changelog: v2.0.0...v3.0.0