v2.3.2

New features

- Added optional calculation of higher order moments when calling
    gmix.get_weighted_moments(obs, with_higher_order=True)
  This is also available with GaussMom(fwhm, with_higher_order=True)

  Currently the sums are calculated and are stored in a larger "sums" and
  "sums_cov" arrays in the results.  No normalized moments are returned.

  Names and indices for moments can be found in moments.MOMENTS_NAME_MAP

Bug Fixes

- Fixed bug when moments are used in guesser and size is bad.
  Only affected rng such as np.random.default_rng that checks
  range for uniform

compatibility

- Convert GMixND.plot to use matplotlib

v2.3.1

See the CHANGES.md for details

v2.3.0

new features

• Caching in pre-psf moments and metacal is now optional with an API to turn it on. Default is off.
• Pixel in pre-PSF moments is not deconvolved when doing PSFs.

v2.2.1

New Features

• Added new guesser BDFPSFFluxGuesser, makes a flux guess
  based on the PSF flux

Performance

• Added O(N) phase shift computation for pre-psf moments

Misc

• ignore NumbaExperimentalFeatureWarning being spewed by
  numba. We can ignore these, as the features are of order
  ten years old from the time of writing

v2.2.0

new features

• Added function to regularize moments results.
• Added numba to pre-PSF moments to reduce runtime.

v2.1.0

tagging v2.1.0

v2.0.6

bug fixes

- Fixed a bug where the `kind` attribute was not set for the `PrePSFMom` fitter.

v2.0.5

v2.0.5

new features

- implemented check for equality, copy() method, copy, and deepcopy
  for Gmix objects, jacobians, and Observation/ObsList/MultiBandObsList
- added get_data() method for jacobian objects
- added apodization to the pre-PSF moments to help prevent FFT artifacts

v2.0.4

Merge pull request #203 from esheldon/beckermr-patch-1

ENH release 2.0.4

v2.0.3

tagging v2.0.3