Transform dimensions are changed when writing to zarr store

Hey everyone,

When I use a scale transform and write/read my data to a zarr store, the dimensions of the scale transform are changed for the image only. That breaks followup aggregations, as the dimensions between the transforms mismatch.

import spatialdata as sd
from spatialdata.datasets import blobs
from spatialdata.transformations import get_transformation, set_transformation, Scale

sdata = blobs()

scale_transform = Scale([0.1, 0.1], ("x", "y"))
set_transformation(sdata["blobs_image"], transformation=scale_transform, to_coordinate_system="global")
set_transformation(sdata["blobs_labels"], transformation=scale_transform, to_coordinate_system="global")

# Initial aggregation works
_ = sdata.aggregate(values="blobs_image", by="blobs_labels", target_coordinate_system="global")

# Write to zarr and read back (as for example recommended for better performance for visualization)
sdata.write("blobs.zarr")
sdata_zarr = sd.read_zarr("blobs.zarr")

# Aggregate no longer works
try:
    _ = sdata_zarr.aggregate(values="blobs_image", by="blobs_labels", target_coordinate_system="global")
except Exception as e:
    print(f"Error: {e}")
# Error: operands could not be broadcast together with shapes (2,) (3,) 

# This is due to a new first dimension added to the scale transform for the image only, resulting in 3 vs 2 dimensions
transform_image = get_transformation(sdata_zarr["blobs_image"], get_all=True)
print(transform_image)
# {'global': Scale (c, y, x)
#    [1.  0.1 0.1]}

transform_labels = get_transformation(sdata_zarr["blobs_labels"], get_all=True)
print(transform_labels)
#{'global': Scale (y, x)
#    [0.1 0.1]}

Start by running the provided reproduction with Scale, sdata.write(), sd.read_zarr(), and aggregate(). Trace how the image and labels transformations are serialized and restored, especially the added image dimension. Done means both transformations retain matching dimensions after the round trip and the post-read aggregation succeeds.