NumPy 2.x compatibility: replace removed np.in1d with np.isin

[NGC2023]

Hi,

With NumPy 2.x, np.in1d has been removed, causing the test suite to fail with:

AttributeError: module 'numpy' has no attribute 'in1d'

The issue can be fixed by replacing np.in1d with np.isin, which provides the same functionality.

--- a/skopt/learning/gbrt.py
+++ b/skopt/learning/gbrt.py
@@ -116,7 +116,7 @@ class GradientBoostingQuantileRegressor(RegressorMixin, BaseEstimator):
         if return_std:
             std_quantiles = [0.16, 0.84]
-            is_present_mask = np.in1d(std_quantiles, self.quantiles)
+            is_present_mask = np.isin(std_quantiles, self.quantiles)
             if not np.all(is_present_mask):
                 raise ValueError(
                     "return_std works only if the quantiles during "

This change restores compatibility with NumPy 2.x while preserving the existing behavior.