Ensure that laplacian and gaussian scaling preserves dtype

[will-moore]

Fixes #495

The gaussian and laplacian scaler functions don't preserve dtype - see issue above.

I looked at tests and there's 2 issues...

• The check_downscaled function

  ome-zarr-py/tests/test_scaler.py

  Line 29 in b49986e

  def check_downscaled(self, downscaled, shape, scale_factor=2):

  only checks that various levels of the pyramid have the same dype as the new 0 level of the same pyramid - but this is NOT the same as the input dtype. In the bug above, ALL levels of the pyramid were float64 when the input was int16 - Fixed this test in the 2nd commit below.

• The tests

  ome-zarr-py/tests/test_scaler.py

  Line 74 in b49986e

  def test_laplacian(self, shape):

  and

  ome-zarr-py/tests/test_scaler.py

  Line 66 in b49986e

  def test_gaussian(self, shape):

  were marked as "fail" because the test asserts that pyramids ONLY get downsampled in X and Y, whereas both these methods downsample in ALL dimensions. - I removed the xfail flag. These 2 tests check dtype now, but not shape.

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 87.04%. Comparing base (b49986e) to head (3d8904b).

Additional details and impacted files

@@            Coverage Diff             @@
##           master     #496      +/-   ##
==========================================
+ Coverage   87.02%   87.04%   +0.02%     
==========================================
  Files          13       13              
  Lines        1772     1776       +4     
==========================================
+ Hits         1542     1546       +4     
  Misses        230      230              

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

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.