Mdciao

A bug regarding example datasets should be fixed
Other release notes:
* Flareplots can have "aura"
* Ions and waters are handled better in the fragmentation heuristics

v2b1
Release candidate

[relase 0.0.8 was yanked because of small bug, we repeat the release notes here]
Most important changes:

* Python 3.12 support added (see note on installation)
* Consensus labels can now be used to streamline analysis across different sequences
* to define neighborhoods
* to define sequence-independent sites
* to refine interface selections (see below)
* as plot labels and as "sort_by" in plots

* API BREAKs
* Definition of interfaces is made easier with refactor:
* `frag_idxs_group_1, frag_idxs_group_2'` become `interface_selection_1, interface_selection_2`
* new parameter `AA_selection`, accepts flexible amino-acid selection
* `self_interface` operates differently, simply deleting self-contacts if not set to True
* Some added returned values of public methods (e.g. `repframes` and `flareplot_attrs`) might break existing code, but its easy to adapt
* One new notebooks and new example datasets
* See the full log for all changes

Relevant PRs
* [python-package] add py312 by gph82 in https://github.com/gph82/mdciao/pull/50
* Select by grn by gph82 in https://github.com/gph82/mdciao/pull/51
* [nomenclature._sort_all_consensus_labels] API CHANGE return also sort… by gph82 in https://github.com/gph82/mdciao/pull/52


**Full Changelog**: https://github.com/gph82/mdciao/compare/0.0.7...0.0.9

Some highlights over the ~ 700 commits
* The neighborlist is not static anymore. All data is screened in a two pass approach: first using pairs of residue-COMs + residue radii and then with actual atom-to-atom distances. This means that the old parameters `interface_cutoff_Ang` and `nlist_cutoff_Ang` are gone, and the user doesn't need to worry about "likely contacts" to speed up the computaton. Theres a number of new methods for this that can be used independently in the `contacts` and the `COM submodules.
* Parallel computing gets a working progressbar-like report
* Methods for making new ContactGroup out of existing ones, either by selecting:
* by [time frames](https://proteinformatics.uni-leipzig.de/mdciao/api/generated/generated/mdciao.contacts.ContactGroup.html#mdciao.contacts.ContactGroup.select_by_frames)
* by [residue (name or index)](https://proteinformatics.uni-leipzig.de/mdciao/api/generated/generated/mdciao.contacts.ContactGroup.html#mdciao.contacts.ContactGroup.select_by_residues)
* by [trajectory](https://proteinformatics.uni-leipzig.de/mdciao/api/generated/generated/mdciao.contacts.ContactGroup.html#mdciao.contacts.ContactGroup.to_ContactGroups_per_traj)

allowing for finer-grained comparisons, e.g. between trajectories or between the beginning and the end the same trajectory.
* Access to KLIFS database can be done via UniProt Accession Code, kinase ID or structure ID
* New time-dependent plotting method [ContactGroup.plot_timedep_ctcs_matrix](https://proteinformatics.uni-leipzig.de/mdciao/api/generated/generated/mdciao.contacts.ContactGroup.html#mdciao.contacts.ContactGroup.plot_timedep_ctcs_matrix), which allows for representations like [this](https://github.com/gph82/mdciao/blob/master/doc/imgs/timedep_ctc_matrix.png).
* The default cutoff is now 4 Å.

Some of the above changes were long overdue TODOs, many other changes, like smaller improvements (API calls have been simplified and harmonized across related methods, still some way to go) are spread throughout the 700 (!) commits.

This patches 0.0.5, the release for the [original manuscript](https://www.biorxiv.org/content/10.1101/2022.07.15.500163v1), against deprecations that happened in the meantime, mostly in numpy handling of arrays, but also in pandas and biopython. Also updated _md_compute_contacts.py to its newest version. There's also smaller changes

Release associated with the submission of mdciao's manuscript

This is mdciao's first stable release candidate