Keep polygon interior during vectorization

There is a minor issue in spatialdata._core.operations.vectorize._vectorize_mask

The interior is removed when we have a mask with an interior, for instance this one:

import numpy as np

mask = np.zeros((100, 100), dtype=np.uint8)
mask[10:90, 10:90] = 1
mask[40:60, 40:60] = 0
mask[45:55, 45:55] = 1

Instead, it will create three polygons nested inside each other. A good behavior should be one of the following:

• one polygon with an interior, and another polygon (for the small square)
• or one multipolygon composed of the two above polygons

I'm not sure if it's very easy to fix. I'll investigate later.

Start in spatialdata._core.operations.vectorize._vectorize_mask and reproduce the issue with the nested-square NumPy mask shown in the report. Inspect the vectorization result and its polygon relationships; done means the outer polygon retains its interior while the small square remains a separate polygon or part of a multipolygon.