Fix casting uint32 to vispy dtype for image layers

[Czaki]

References and relevant issues

extracted from #6439

Description

Before this PR

After this PR

Script to reproduce:

import numpy as np

import napari

# Set the number of steps
nb_steps = 2**17

base = np.linspace(start=1, stop=nb_steps, num=nb_steps).astype('uint32')
label_img = np.repeat(
        base.reshape([1, base.shape[0]]), int(nb_steps/1000), axis=0
        )

viewer = napari.Viewer()
viewer.add_image(label_img, colormap='viridis', contrast_limits=(0, nb_steps), scale=(100, 1))

napari.run()

[codecov]

Codecov Report

All modified and coverable lines are covered by tests ✅

Comparison is base (2fc6124) 92.23% compared to head (0a34176) 92.16%.

Additional details and impacted files

@@            Coverage Diff             @@
##             main    #6456      +/-   ##
==========================================
- Coverage   92.23%   92.16%   -0.07%     
==========================================
  Files         601      601              
  Lines       53167    53174       +7     
==========================================
- Hits        49037    49009      -28     
- Misses       4130     4165      +35     

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

[jni]

@Czaki what do you think about adding the script to examples/dev or examples/dev/issues? We could make a habit of that because sometimes our tests don't capture the final behavior that we see in the canvas (see #6411 breaking direct color mode). So it would be nice to have a set of scripts that we can go back to to figure out when something untested broke.