`render_points` with column coloring throws error when there are too many points

Recommendation: attach a minimal working example

Minimal working example with a toy dataset:

from spatialdata.models import PointsModel
n_points=10000
coords = pd.DataFrame({'x': np.random.rand(n_points), 
                       'y': np.random.rand(n_points)})
points = PointsModel.parse(coords)
points_sdata = sd.SpatialData(points={"points": points})

obs = pd.DataFrame()
obs["instance_id"] = np.arange(n_points)
obs["region"] = "points"
obs["region"].astype("category")
obs["feature"] = np.random.rand(n_points)

table = AnnData(X=np.random.rand(n_points, 1), obs=obs)

points_sdata['table'] = TableModel.parse(table, region='points', region_key='region', instance_key='instance_id')

points_sdata.pl.render_points('points', color='feature', size=100).pl.show()

Describe the bug
Changing n_points from 10000 to 10001 and greater values gives the error: KeyError: 'feature' which is very annoying because I work with very large single cell datasets.

My assessment:
looking at this source code

if method is None:
        method = "datashader" if len(points) > 10000 else "matplotlib"

it seems that the problem is due to using the datashader backend?

Start at the render_points entry point and inspect the automatic method selection around the 10,000-point threshold, reproducing the supplied example with 10,000 and 10,001 points. Trace how the feature column is handled by the large-point rendering path; done means column coloring works without KeyError for datasets above the threshold.