Add isort to pre-commit hooks, package resorting

[charlesbluca]

Second pass at #4600, as there were a lot of merge conflicts piling up there.

• Tests added / passed
• Passes black distributed / flake8 distributed

[charlesbluca]

Looks like failures due to deprecations in pytest - does this have something to do with #4646?

[jrbourbeau]

Yep, I was trying to figure out what's causing the test suite to start failing (definitely unrelated to the changes here). It looks like there was an issue in a new release of the setup-miniconda GitHub action (xref conda-incubator/setup-miniconda#160), though this should be fixed soon for us with #4585

[charlesbluca]

Good to know! In that case, I'll revisit the tests once that is merged.

[jrbourbeau]

#4585 is in, so you should be able to resolve the test failures here by merging main into your PR branch

[jrbourbeau]

Thanks @charlesbluca!

I have a similar question to this one over in dask/dask about making dask a special third party library so it's imports are just above the distributed imports instead of being grouped with things like numpy, tlz, etc.

cc @jsignell

[jrbourbeau]

It looks like release came out a few days ago

Suggested change

	rev: 5.7.0
	rev: 5.8.0

[jrbourbeau]

Could you add a small comment here about why we need to skip isort?

[charlesbluca]

Sure!

[charlesbluca]

Thanks for the review! Distinguishing Dask from other imports makes sense to me, happy to add that in as long as it's consistent with dask/dask#7370.

[jrbourbeau]

We just added a new module which is throwing some isort errors. Would you mind running isort on the new module and pushing up the changes?

[charlesbluca]

Sure! Also want to note that a test that failed in the original PR #4600, test_as_completed_with_results, seems to have worked in these subsequent CI runs, so maybe it should be marked as flaky.

[charlesbluca]

Most of the test failures have something to do with human readable byte formatting not being consistent:

 >           assert "GB" in text and "4" in text
E           assert ('GB' in "LocalCluster(dd8123bb, 'inproc://10.1.0.4/2069/705', workers=2, threads=4, memory=3.73 GiB)")

Not sure if this has something to do with the order of imports, as I'm unable to replicate these failures locally.

[jsignell]

Most of the test failures have something to do with human readable byte formatting not being consistent:

This has to do with the work that @crusaderky has been doing in dask/dask#7484 it'll be fixed in #4649

[jrbourbeau]

Thanks @charlesbluca!

[charlesbluca]

Thanks for the help @jrbourbeau and @jsignell!