Fix gen_cluster failures; dask_version tweaks

[crusaderky]

• fixes one of the issues reported in ⚠️ Nightly upstream-dev CI failed ⚠️ #5600
• distributed.utils_test.gen_cluster no longer accepts timeout=None for the sake of robustness
• deleted ancient dask backwards compatibility code
• clean up code around dask.__version__

[pep8speaks]

Hello @crusaderky! Thanks for updating this PR. We checked the lines you've touched for PEP 8 issues, and found:

There are currently no PEP 8 issues detected in this Pull Request. Cheers! 🍻

Comment last updated at 2021-07-15 16:43:47 UTC

[github-actions]

Unit Test Results

         6 files  ±0           6 suites  ±0   54m 25s ⏱️ ±0s
16 171 tests ±0  14 440 ✔️ ±0  1 730 💤 ±0  1 ❌ ±0 
90 222 runs  ±0  82 071 ✔️ ±0  8 145 💤 ±0  6 ❌ ±0 

For more details on these failures, see this check.

Results for commit bc92331. ± Comparison against base commit bc92331.

♻️ This comment has been updated with latest results.

[dcherian]

You'll need a commit with [test-upstream] in the message to run the upstream-dev CI

[keewis]

thanks for the quick fix and the clean-up, @crusaderky.

In case of temporary errors (such as the mamba connection errors that sometimes appear) there is a "Rerun all jobs" button to rerun the workflow (it needs to either be cancelled or completed first, though). Closing and reopening works, too, but it reruns everything (including RTD) and the list of checks is extended.

[keewis]

the remaining error is the other one reported in #5600 so this should be ready to merge. Unless you want to add a "internals" whats-new.rst entry?

[crusaderky]

Unless you want to add a "internals" whats-new.rst entry?

I think it may be a bit overkill?