Update scikit-learn version requirement to support v1.6.0

[marta-guimaraes]

Hello!

I’ve encountered a compatibility issue with the current install_requires setting in the setup.py file:

install_requires=[
    ...
    "scikit-learn>=1.0.1,<1.6.0",
]

The upper limit (<1.6.0) prevents the use of the latest stable version of scikit-learn (1.6.0). This creates conflicts in projects where scikit-learn is already updated to the latest version.

---

Note:
Previously, the function sklearn.metrics.mean_squared_error allowed the RMSE calculation with the argument squared=True. However, in scikit-learn 1.6.0, the squared argument has been removed. Instead, a dedicated function, sklearn.metrics.root_mean_squared_error, has been introduced.

Admittedly, such a breaking change in a minor version release (1.6.0) was unexpected, as this would typically be reserved for a major version update. Regardless, it impacts the compatibility between evidently and scikit-learn in its latest release.