Use copy of x0 in scipy.optimize.newton to prevent side effect

[maierbn]

This fixes the linked issue.

[johannes-mueller]

Is the new test actually sensitive for the change? I reverted the production code changes locally and the new test still passes.

[maierbn]

I'm not sure, maybe it has to do with python 3.13?

[johannes-mueller]

Or maybe Windows <> Linux?

So it does fail on your system without the changes?

[maierbn]

No, the issue does not occur for me in the test. But when do basically the same in a different package that imports pylife (but in python 3.13), the issue occurs. Is it required to have a test for this PR to be mergable?

[johannes-mueller]

Ah ok. There is actually a lot more issues with python 3.13 that I intent to address. Maybe check this workflow here

https://github.com/boschresearch/pylife/actions/runs/11832914483/job/32970554847

Maybe we should do that in one PR and at the same time add a python-3.13 test in the usual test suite.