fix: statisitcal map centroid calculation resulting in NaN points

siibra/volumes/parcellationmap.py

@@ -593,7 +593,7 @@ def compute_centroids(self) -> Dict[str, point.Point]:
             with QUIET:
                 mapimg = self.fetch(index=index)  # returns a mask of the region
             maparr = np.asanyarray(mapimg.dataobj)
-            centroid_vox = np.mean(np.where(maparr == 1), axis=1)
+            centroid_vox = np.mean(np.nonzero(maparr), axis=1)
             assert regionname not in centroids
             centroids[regionname] = point.Point(
                 np.dot(mapimg.affine, np.r_[centroid_vox, 1])[:3], space=self.space