1.2 Release

[Marius1311]

to do's before this release

• [ ] finish pyGPCCA integration, make sure tests pass (#472)
• [ ] close the legend plotting PR (#470)
• [ ] write release notes

I will suggest text for the release notes below.

[Marius1311]

Suggested release notes:

1.2.0

This release includes:

Bugfixes

• fixed a bug when computing the Schur decomposition for reducible Markov chains (Schur vectors appear to not be D-orthogonal.). GPCCA requires the leading Schur vectors to be orthogonal wrt. a symmetric, positive definite matrix D. [#453]
• @michalk8, add description of #474 and potentially other bugfixes you did

Additions

• Switched from Travis CI to GitHub actions due to pricing changes in Travis CI. This ensures CellRank remains easy to maintain and extend. [#452]
• Completely refactored the underlying code base of GPCCA and set it up as it's own package called pyGPCCA with documentation and comprehensive continuous integration via GitHub Actions. Going forwards, this ensures that one of the "engines" of CellRank is also easy to maintain to extend. Further, this will make CellRank's installation more convenient by not needing to vendorize any longer.
• Added support of circular embeddings visualising computed fate probabilities as it is done in STEMNET (@michalk8, insert citation here) [#457]
• Allowing legends not to be plotted by passing legend_loc="none", as it is done in scvelo[#470]