fix: make_glm_data ordinal crash on NumPy 2.x

np.random.choice(..., 1, p=...) with size=1 returns a 1-element array
on NumPy 2.x, causing ValueError when assigning to y[i]. Drop the size
argument so it returns a scalar on all NumPy versions.

Generated with [Claude Code](https://claude.ai/code)
via [Happy](https://happy.engineering)

Co-Authored-By: Claude <noreply@anthropic.com>
Co-Authored-By: Happy <yesreply@happy.engineering>

Author: 3 people

python/abess/datasets.py

@@ -314,7 +314,7 @@ def __init__(self, n, p, k, family, rho=0, corr_type="const", sigma=1,
             # y
             y = np.zeros(n)
             for i in range(n):
-                y[i] = np.random.choice(np.arange(class_num), 1, p=prob[i, :])
+                y[i] = np.random.choice(np.arange(class_num), p=prob[i, :])
         else:
             raise ValueError(
                 "Family should be \'gaussian\', \'binomial\', "