Ultranest

------------------

* new :py:class:`ultranest.mlfriends.LocalAffineLayer` for metric learning, set as default (see `issue 124 <https://github.com/JohannesBuchner/UltraNest/issues/124>`_)
* add Highest Density Interval function (ultranest.plot.highest_density_interval_from_samples)
* corner plot style with higher signal-to-ink ratio.
* bug fixes in popstepsampler

------------------

* add number of steps calibrator :py:class:`ultranest.calibrator.ReactiveNestedCalibrator`
* add relative jump distance diagnostic for step samplers
* make population step samplers more consistent with other step samplers

------------------

* new :py:class:`ultranest.mlfriends.MaxPrincipleGapAffineLayer` for metric learning, set as default

------------------

* documentation improvements
* logging with MPI fixes `by adipol-ph <https://github.com/JohannesBuchner/UltraNest/issues/109>`_ and `by gregorydavidmartinez <https://github.com/JohannesBuchner/UltraNest/issues/110>`_
* more flexible plotting `by facero <https://github.com/JohannesBuchner/UltraNest/issues/108>`_

------------------

* add PopulationRandomWalkSampler: vectorized Gaussian random walks for GPU/JAX-powered likelihoods
* limit initial widening to escape plateau (issue 81)

------------------

* add hot-resume: resume from a similar fit (with different data)
* fix post_summary.csv column order
* fix build handling for non-pip systems (pyproject.toml)
* more efficient handling of categorical variables