Use integer division for new_shape in resize_image

[will-moore]

Fixes issue reported at #219 (comment)

The issue there was that the dask writing used resize_image() which calculated down-sizing using ceil() so the sizes of pyramids levels were 100, 50, 25, 13, 7 instead of 100, 50, 25, 12, 6 as for the non-dask data.

I updated resize_image() to use the same integer division for calculating new_shape as is used in the def __nearest() for non-dask data.

To test:

• run the sample script on issue above (using ome-zarr-py installed from this branch)
• check that the pyramid sizes are 100, 50, 25, 12, 6 for both the debug0.zarr (non-dask) and debug1.zarr (dask).
• The coordinateTransforms should also be the same for both

[codecov]

Codecov Report

Base: 84.26% // Head: 84.46% // Increases project coverage by +0.20% 🎉

Coverage data is based on head (a1fdcf5) compared to base (0c1b1bb).
Patch coverage: 100.00% of modified lines in pull request are covered.

Additional details and impacted files

@@            Coverage Diff             @@
##           master     #220      +/-   ##
==========================================
+ Coverage   84.26%   84.46%   +0.20%     
==========================================
  Files          13       13              
  Lines        1474     1474              
==========================================
+ Hits         1242     1245       +3     
+ Misses        232      229       -3     

Impacted Files	Coverage Δ
ome_zarr/scale.py	68.42% <100.00%> (ø)
ome_zarr/reader.py	86.05% <0.00%> (+0.78%)	⬆️

Help us with your feedback. Take ten seconds to tell us how you rate us. Have a feature suggestion? Share it here.

☔ View full report at Codecov.
📢 Do you have feedback about the report comment? Let us know in this issue.

[joshmoore]

The change & explanation make sense, @will-moore. I'd strongly suggest though that we get the test migrated over from the issue.

[will-moore]

@joshmoore Added more tests, and with #221 they're green!

[joshmoore]

Thanks, @will-moore.