version 1.3.0

This release involves some major behind-the-scenes changes to improve performance.

Major changes

• move to only storing weight > 0 pixels in observations, by default. This greatly improves performance and memory usage when fitting heavily masked images
• added leastsqbounds to deal with bounded parameters in LM. This is supported by adding bounds to priors.

Added Features

• added bulge+disk class and fitting. This has no constraint on the ratio of Tdev to Texp (TdByTe)
• added Sinh prior

Breaking changes

• TdByTe can be sent to the GMixBDF class, and bdf is no longer supported with GMixModel

v0.9.0

First official release