Temporaray pin of scipy version

[AVHopp]

This issue serves as a reminder that we currently need to explicitly pin the scipy version to <1.15.

This is necessary as this new version caused several problems in other packages, including Botorch and is currently also being pinned there (see here). We can probably remove the explicit pin on our side again once the next version of Botorch has been released, hence the "on hold" label.

[AVHopp]

/remind me to check for a new Botorch version next month.

[github-actions]

@AVHopp set a reminder for 2/14/2025

[Scienfitz]

It appears the new botorch version is out, according to the CHANGELOG something was done to fix computational problems with newer scipy versions

There are also other changes relevant to us:

• Require GPyTorch==1.14: Likely the cause of the new kernel test failure @AdrianSosic
• new qPosteriorStandardDeviation which enables batch optimization for active learning with this cheap variant (qNIPV can often be too expensive). I will include this acqf during fixing Inconsistent use of is_mc property #467
• InfeasibilityError relevant for Botorch with cardinality constraint via sampling #301 and @Waschenbacher

[Scienfitz]

cant find any issue anymore:

(ignore error state its unrelated to this issue here)
Should we unpin scipy and require botorch>=0.13.0 ?