"batch_size" parameter doesn't work for "sc.external.pp.scanorama_integrate"

According to scanorama MemoryError, I think batch_size parameter doesn't work for sc.external.pp.scanorama_integrate

Minimal code sample (that we can copy&paste without having any data)

sc.external.pp.scanorama_integrate(adata, key="datasetID", batch_size=500)

WARNING: Out of memory, consider turning on batched computation with batch_size parameter.
Traceback (most recent call last):
  File "/lustre1/shiq//02_igt/py/1_igt_scanorama.py", line 26, in <module>
    sc.external.pp.scanorama_integrate(adata, key="datasetID", batch_size=500)
  File "/lustre1/shiq/app/miniconda3/envs/scanpy/lib/python3.10/site-packages/scanpy/external/pp/_scanorama_integrate.py", line 121, in scanorama_integrate
    integrated = scanorama.assemble(
  File "/lustre1/shiq/app/miniconda3/envs/scanpy/lib/python3.10/site-packages/scanorama/scanorama.py", line 933, in assemble
    bias = transform(curr_ds, curr_ref, ds_ind, ref_ind, sigma=sigma,
  File "/lustre1/shiq/app/miniconda3/envs/scanpy/lib/python3.10/site-packages/scanorama/scanorama.py", line 762, in transform
    avg_bias = batch_bias(curr_ds, match_ds, bias, sigma=sigma,
  File "/lustre1/shiq/app/miniconda3/envs/scanpy/lib/python3.10/site-packages/scanorama/scanorama.py", line 723, in batch_bias
    weights = rbf_kernel(curr_ds, match_ds, gamma=0.5*sigma)
  File "/lustre1/shiq/app/miniconda3/envs/scanpy/lib/python3.10/site-packages/sklearn/metrics/pairwise.py", line 1294, in rbf_kernel
    X, Y = check_pairwise_arrays(X, Y)
  File "/lustre1/shiq/app/miniconda3/envs/scanpy/lib/python3.10/site-packages/sklearn/metrics/pairwise.py", line 155, in check_pairwise_arrays
    X = check_array(
  File "/lustre1/shiq/app/miniconda3/envs/scanpy/lib/python3.10/site-packages/sklearn/utils/validation.py", line 727, in check_array
    warnings.warn(
FutureWarning: np.matrix usage is deprecated in 1.0 and will raise a TypeError in 1.2. Please convert to a numpy array with np.asarray. For more information see: https://numpy.org/doc/stable/reference/generated/numpy.matrix.html

Read scanpy/external/pp/_scanorama_integrate.py, especially the scanorama_integrate entry point and its scanorama.assemble call, then trace batch_size into the dependency’s assemble/transform path. Reproduce the provided call and traceback; done means the parameter reaches the intended batched computation without the reported out-of-memory path.