fix: use np.array instead of np.asarray to ensure writable array

np.asarray returns a read-only view when the input dtype already matches,
causing np.nan_to_num(copy=False) to raise ValueError on numpy 1.25+ with
Python 3.11-3.13. np.array always produces a writable copy.

Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>

Author: aeiwz

metbit/analysis/opls_da.py

@@ -198,7 +198,7 @@ def __init__(
             _dtype = dtype
 
         # Single allocation: extract to numpy with target dtype, replace NaN in-place.
-        X_arr = X.to_numpy(dtype=_dtype) if isinstance(X, pd.DataFrame) else np.asarray(X, dtype=_dtype)
+        X_arr = X.to_numpy(dtype=_dtype) if isinstance(X, pd.DataFrame) else np.array(X, dtype=_dtype)
         np.nan_to_num(X_arr, copy=False)
         X_df = pd.DataFrame(X_arr, columns=resolved_features)