Arviz

New features
- Add optimized simultaneous ECDF confidence bands ([2368](https://github.com/arviz-devs/arviz/pull/2368))
- Add support for setting groups with `idata[group]` ([2374](https://github.com/arviz-devs/arviz/pull/2374))

Maintenance and fixes
- Make `dm-tree` and optional dependency ([2379](https://github.com/arviz-devs/arviz/pull/2379))
- Fix bug in `psislw` modifying input inplace ([2377](https://github.com/arviz-devs/arviz/pull/2377))
- Fix behaviour of two dimensional KDE plot with recent matplotlib releases ([2383](https://github.com/arviz-devs/arviz/pull/2383))
- Make defaults in `plot_compare` more intuitive ([2388](https://github.com/arviz-devs/arviz/pull/2388))

Documentation
- Added extensions of virtual environments in [.gitignore](https://github.com/arviz-devs/arviz/blob/main/.gitignore) ([2371](https://github.com/arviz-devs/arviz/issues/2371))
- Fixed the issue in the [Contribution References Documentation](https://python.arviz.org/en/latest/contributing/index.html) ([2369](https://github.com/arviz-devs/arviz/issues/2369))
- Improve docstrings for `loo` and `waic` ([2366](https://github.com/arviz-devs/arviz/pull/2366))

New features
- Use revised Pareto k threshold ([2349](https://github.com/arviz-devs/arviz/pull/2349))
- Added arguments `ci_prob`, `eval_points`, `rvs`, and `random_state` to `plot_ecdf` ([2316](https://github.com/arviz-devs/arviz/pull/2316))
- Deprecated rcParam `stats.hdi_prob` and replaced with `stats.ci_prob` ([2316](https://github.com/arviz-devs/arviz/pull/2316))
- Expose features from [arviz-base](https://arviz-base.readthedocs.io), [arviz-stats](https://arviz-stats.readthedocs.io) and [arviz-plots](https://arviz-plots.readthedocs.io) as `arviz.preview`
submodule ([2361](https://github.com/arviz-devs/arviz/pull/2361))

Maintenance and fixes
- Ensure support with numpy 2.0 ([2321](https://github.com/arviz-devs/arviz/pull/2321))
- Update testing strategy to include an environment without optional dependencies and
an environment with [scientific python nightlies](https://anaconda.org/scientific-python-nightly-wheels) ([2321](https://github.com/arviz-devs/arviz/pull/2321))
- Address bokeh related deprecations ([2362](https://github.com/arviz-devs/arviz/pull/2362))

- Fix legend overwriting issue in `plot_trace` ([2334](https://github.com/arviz-devs/arviz/pull/2334))

Deprecation
- Support for arrays and DataArrays in plot_khat has been deprecated. Only ELPDdata will be supported in the future ([2349](https://github.com/arviz-devs/arviz/pull/2349))
- Removed arguments `values2`, `fpr`, `pointwise`, and `pit` in `plot_ecdf` ([2316](https://github.com/arviz-devs/arviz/pull/2316))

New features
- Add new example data `rugby_field` and update `rugby` example data ([2322](https://github.com/arviz-devs/arviz/pull/2322))
- Support for `pytree`s and robust to nested dictionaries. ([2291](https://github.com/arviz-devs/arviz/pull/2291))
- Add `.close` method to `InferenceData` ([2338](https://github.com/arviz-devs/arviz/pull/2338))


Maintenance and fixes
- Fix deprecation warnings in multiple dependencies ([2329](https://github.com/arviz-devs/arviz/pull/2329),
[2332](https://github.com/arviz-devs/arviz/pull/2332) and [2333](https://github.com/arviz-devs/arviz/pull/2333))

Deprecation

- Removed arguments `values2`, `fpr`, `pointwise`, `npoints`, and `pit` in `plot_ecdf` ([2316](https://github.com/arviz-devs/arviz/pull/2316))

Documentation

Maintenance and fixes
- Fix deprecations introduced in latest pandas and xarray versions, and prepare for numpy 2.0 ones ([2315](https://github.com/arviz-devs/arviz/pull/2315)))
- Refactor ECDF code ([2311](https://github.com/arviz-devs/arviz/pull/2311))
- Fix `plot_forest` when Numba is installed ([2319](https://github.com/arviz-devs/arviz/pull/2319))

New features
- Add prior sensitivity diagnostic `psens` ([2093](https://github.com/arviz-devs/arviz/pull/2093))
- Add filter_vars functionality to `InfereceData.to_dataframe`method ([2277](https://github.com/arviz-devs/arviz/pull/2277))

Maintenance and fixes

- Update requirements: matplotlib>=3.5, pandas>=1.4.0, numpy>=1.22.0 ([2280](https://github.com/arviz-devs/arviz/pull/2280))
- Fix behaviour of `plot_ppc` when dimension order isn't `chain, draw, ...` ([2283](https://github.com/arviz-devs/arviz/pull/2283))
- Avoid repeating the variable name in `plot_ppc`, `plot_bpv`, `plot_loo_pit`... when repeated. ([2283](https://github.com/arviz-devs/arviz/pull/2283))
- Add support for the latest CmdStanPy. ([2287](https://github.com/arviz-devs/arviz/pull/2287))
- Fix import error on windows due to missing encoding argument ([2300](https://github.com/arviz-devs/arviz/pull/2300))
- Add ``__delitem__`` method to InferenceData ([2292](https://github.com/arviz-devs/arviz/pull/2292))

Documentation
- Improve the docstring of `psislw` ([2300](https://github.com/arviz-devs/arviz/pull/2300))
- Rerun the quickstart and working with InferenceData notebooks ([2300](https://github.com/arviz-devs/arviz/pull/2300))

- Several fixes in `plot_ppc` docstring ([2283](https://github.com/arviz-devs/arviz/pull/2283))

Maintenance and fixes

- Fix Numba deprecation errors and incorrect nopython usage ([2268](https://github.com/arviz-devs/arviz/pull/2268))

Documentation

- Rerun Numba notebook