chore: remove upper bound on numpy and scipy versions

Remove the upper version constraints on numpy (<2) and scipy (<1.12.0)
to allow compatibility with numpy 2.x and newer scipy versions.

- numpy: Changed from ">=1.26.0, <2" to ">=1.26.0"
- scipy: Changed from ">=1.7.3, <1.12.0" to ">=1.7.3"

This enables users to install giskard alongside other libraries that
require numpy 2.x, which has been stable for over a year.

Note: PR #2222 already fixed the deprecated scipy.stats.stats imports
that were blocking scipy 1.12+ compatibility.

Fixes #2173

Signed-off-by: majiayu000 <[email protected]>

Author: majiayu000

pyproject.toml

@@ -155,9 +155,9 @@ dependencies = [
     "cloudpickle>=1.1.1",
     "zstandard>=0.10.0",
     "mlflow-skinny>=2",
-    "numpy>=1.26.0, <2",
+    "numpy>=1.26.0",
     "scikit-learn>=1.0",
-    "scipy>=1.7.3, <1.12.0",
+    "scipy>=1.7.3",
     "mixpanel>=4.4.0",
     "requests>=2.19",
     "pydantic<3,>1",