Add generation of u_nk and dHdl from linearly related dependencies

* Add moving_average function for visualization and convergence testing

* Update versionadded

* Run Black

* Bug fix bar_.py states

* Update Changelog

* Update the docs

* Add tests

* Formatting to align with Black

* Update tests

* Refactor moving_average to align with forward_backward_convergence function

* Update tests

* Update test_convergence and lambda tests in convergence.moving_average

* Adjust convergence.py and tests for codecoverage

* black

* Update moving_average to block_average for more accurate descriptive name

* Address reviewer comments

* Update test to align with changed handling of dfs of different length in block_average

* Remove incorrect popagation of error in BAR

* Add tests and error catch for ill constructed BAR input, u_nk

* black

* Updated version comments

---------

Co-authored-by: Oliver Beckstein <[email protected]>

- generated with https://bit.ly/cffinit
- authors and order from AUTHORS
- affiliations and ORCID from JOSS paper
- emails from paper or individuals

Co-authored-by: David L. Dotson <[email protected]>

- inferred DOI from proof
- assume final acceptance in 2024...

- added David Mobley @davidlmobley (original conception and design of alchemlyb)
- added Michael Shirts @mrshirts (original conception and design of alchemlyb)

- CHANGES: bump to 2.3.3
- add reminder to AUTHORS to also update CITATION.cff

Added all authors that were mentioned in the Acknowledgements of the JOSS
paper but were not in AUTHORS

- Wei-Tse Hsu @wehs7661 (2020) for code clean-up in a732380
- Jan Janssen  @jan-janssen (2022) for creating the conda-forge package
- Shujie Fan @VOD555 (2022) for initial code for fractional equilibration time
- Helmut Carter @helmutcarter (2024) for doc fix in alchemistry#356

AUTHORS entry had been forgotten in PR alchemistry#381

* fix bug introduced in PR alchemistry#381: there was a change to creating the delta_f_ matrix,
  which resulted in the columns and indices being tuples that were in the wrong order 
  for single lambda computations.
* ensure that columns are in the correct order by explicitly sorting
* add a test for the delta_f_ columns

- fix alchemistry#385
- use pyproject.toml instead of setup.py (note: may need to change README for PyPi to exclude
  banners)
- remove versioneer and use versioningit (use alchemlyb.__version__ directly where
  the version is needed, e.g., for sphinx docs)
- updated CHANGES for 2.4.0

- gitignore _version
- ignore commits for blame that contain black-reformatting
  (in particular we forgot to include alchemistry#280)

- close alchemistry#402
- use same title in CITATION.cff for paper and project
  (required for the JOSS paper publication alchemistry#71)
- added DOI for preferred citation in note because that is the
  only hint that shows up on zenodo)

replace `pandas.concat()` with `alchemlyb.concat()` in the tutorial (given that we explicitly tell users to use it)