Clean up tests and reorganize test artifacts

[keithchev]

Context

Before addressing the open PRs, I wanted to improve the test coverage. Before doing that, I had to understand the existing tests, and to do that, I felt it made sense to rewrite the tests for clarity, since they were a bit hard to follow.

Overview

This PR cleans up the tests by rewriting them to improve nomenclature and clarity. It also slightly expands the tests where it was possible to include some trivial additional cases/tests. Some tests were also split into two tests, in cases where different functionality was being tested in the same test.

It also reorganizes the test artifacts to clarify the relationship between some of the LIF files and their expected outputs (which are stored in the form of TIFF files).

Finally, it adds a README to /tests/artifacts with a brief explanation of what each test LIF file contains (as determined by manually inspecting how it is loaded by BioFormats in Fiji).

Note for reviewers

The changes to the tests are so substantial that there is no diff; the tests appear as new files. FWIW, I was careful not to change the substance of the tests or to remove any tests.

[ekiefl]

I liked the README in the artifacts directory and the layout of conftest.py. Looks good to me!

[lanery]

Tests look great! I left a few questions/comments for the artifacts/README.

The changes to the tests are so substantial that there is no diff; the tests appear as new files.

No diff bc the name changed from tests/test_readlif.py --> tests/test_reader.py.

[lanery]

I think it's T3.

Side note: the channel dimension seems to be missing from Dims?

>>> lif_file.image_list
[{'dims': Dims(x=1024, y=1024, z=3, t=3, m=1),
  'display_dims': (1, 2),
  'dims_n': {1: 1024, 2: 1024, 3: 3, 4: 3},
  'scale_n': {1: 9.8709062997224,
   2: 9.8709062997224,
   3: 2.9789060682401844,
   4: 0.19273393080851883},
  'path': 'xyzt test/',
  'name': 'Series002',
  'channels': 2,
  'scale': (9.8709062997224,
   9.8709062997224,
   2.9789060682401844,
   0.19273393080851883),
  'bit_depth': (8, 8),
  'mosaic_position': [],
  'channel_as_second_dim': False,
  'settings': {}}]

[keithchev]

Yeah, the dims attribute contains all of the dimensions other than the channel dimension, which appears in the separate channels attribute. I think this is because the channels are stored separately in the LIF metadata, but I'm not sure.

[keithchev]

Also, dims only contains XYZTM, so it doesn't include the lambda-scan-related dimensions (excitation and emission wavelengths).