Fix tests

[timholy]

The random seed is not stable across Julia versions, so the random
matrices are not consistent. On Julia 1.10 (and some other versions too),
the random matrix in sparse_test is not full rank, so the test fails.
This adds a small diagonal to ensure it has full rank.

[JeffFessler]

@ahwillia would you mind adding @timholy as a maintainer of this package so that he can do this work more easily? I can approve his workflow runs but I do not have privileges to add maintainers.

Perhaps ideally this package would get put in some organization where there might be a set of possible maintainers, but I am unsure which of the many optimization-related orgs would fit it best.

[JeffFessler]

Seems to be some error in the pycall.

[ahwillia]

Done. Also happy to transfer the package somewhere else if helpful.

[timholy]

Maybe https://github.com/JuliaLinearAlgebra?

[JeffFessler]

That org makes sense because it already has other solvers like:
https://github.com/JuliaLinearAlgebra/IterativeSolvers.jl

[timholy]

With regard to the scipy failure, to me it looks like we're being bitten by scipy/scipy#21140. As that will clearly resolve itself (see scipy/scipy#21021), and tests pass for me locally, I think we should just ignore it for now.

[ahwillia]

@timholy, I tried to initiate the transfer but gives me an error message below. Do I need to be added to JuliaLinearAlgebra temporarily to make the transfer? Or is there a better way?

You don’t have the permission to create public repositories on JuliaLinearAlgebra

[timholy]

I think you'll need an invitation. I have that power 💪 . As a courtesy to the other owners of JuliaLinearAlgebra, I posted our plans on the slack #linear-algebra channel. It's possible someone will suggest an even better org. But if I haven't heard any objections by the end of today, I'll send you an invite to JuliaLinearAlgebra.

[timholy]

Invitation sent.