Pass the polygon data to `self._viewer.add_shapes()` in a more efficient way

Right now we have the polygons data inside a geodataframe object, and we convert this to a list of list of floats to pass it to self._viewer.add_shapes() with this code

coordinates = polygons.geometry.apply(lambda x: np.array(x.exterior.coords).tolist()).tolist()

which takes 4 seconds for 160K polygons.

We have to check:

• if there is a better way to access this information within the GeoDataFrame object (maybe some buffer). I don't think this is possible
• if there is a better way to pass this data to napari, maybe the ragged array representation.
• if we can parallelize the conversion of this data.

Start by profiling the shown polygons.geometry.apply(...).tolist() conversion and its call to self._viewer.add_shapes(); the issue names no files or tests. Compare GeoDataFrame access, napari ragged-array input, and parallelization, then benchmark the chosen approach against 160K polygons. Done means a measured improvement with equivalent polygon data rendered correctly.