[bug] tests on Mac OS time out

[jan-janssen]

Locally they seem to work fine.

[jan-janssen]

It is strange that the version of the Mac OS containers was downgraded:

12.7
21G816

vs.

 macOS
 12.6.9
 21G726

[jan-janssen]

A similar issue was already reported in actions/runner-images#8642

[jan-janssen]

Currently effected pull requests:

• Bump mpi4py from 3.1.4 to 3.1.5 #181
• Split Slurm and MPI interface #190
• Fix interface shutdown #193

[liamhuber]

From the CI on #240:

It tentatively looks like the hanging might involve the OSX version, but since 13 is also hanging I'm not convinced just getting a more-up-to-date 12 is going to solve things. 11 is working fine.

[liamhuber]

As discussed in the meeting today, OSX tests run fine on bare metal for two of the versions in #240, including failures. The conclusion from this is that the problem is likely MPI+the CI runners, and not pympipool+OSX.

The long-term solution is still to hope that the generic issue @jan-janssen linked above gets resolved.

In the meantime, OSX11 is working extremely consistently, so a stop-gap solution is to use that instead of macos-laters. I plan to go over to pyiron/actions and make the OS versions a workflow variable. In the meantime I also downgraded macos in #241.

IMO this should still be left open as long as macos-latest fails.

[jan-janssen]

Fixed in #258