Implement the Jordan decomposition of a matrix and is_semisimple() test

[tscrim]

We implement the Jordan decomposition of a matrix $A = S + N$, where $S$ is a semisimple matrix and $N$ is a nilpotent matrix. We also implement a check for the matrix being semisimple by using the minpoly() being square-free.

📝 Checklist

• The title is concise and informative.
• The description explains in detail what this PR is about.
• I have linked a relevant issue or discussion.
• I have created tests covering the changes.
• I have updated the documentation accordingly.

⌛ Dependencies

• Implement an is_nilpotent() method for matrices #37703 For a doctest about the decomposition

[mkoeppe]

LGTM.

For inexact fields, that while loop looks a bit scary, but that can be the concern of specialized implementations.

[tscrim]

Thanks.

Indeed, that is a good point about the inexact fields. I have now forbidden it. I also made the checks a bit better for things over nonfields.

Actually, I found something that could actually make this a bit better even:

https://acl.inf.ethz.ch/people/chrisw/projects/felix/thesis.pdf

[tscrim]

This seems to be the best possible algorithm. Sorry for a bit of noise.

[mkoeppe]

Nice, thanks for working on this improvement.

[vbraun]

merge conflict

[tscrim]

Rebased (although I found no merge conflict).

[github-actions]

Documentation preview for this PR (built with commit 70809e9; changes) is ready! 🎉