Offline testing and test data fetching improvements

[Zeitsperre]

Pull Request Checklist:

• This PR addresses an already opened issue (for bug fixes / features)
  • This PR fixes Test suite does not run without internet access #1468
• Tests for the changes have been added (for bug fixes / features)
  • (If applicable) Documentation has been added / updated (for bug fixes / features)
• CHANGES.rst has been updated (with summary of main changes)
  • Link to issue (:issue:number) and pull request (:pull:number) has been added

What kind of change does this PR introduce?

• Adds pytest-socket to the testing dependencies, in order to block HTTP requests
• Adds a new pytest marker "requires_internet" for tests that absolutely depend on internet access.
• Fixed an error in the pytest configuration that impeded copying of testing data to thread safe caches of workers (this should always occur).
• Added handling for pytest-socket's SocketBlockedError when attempting to fetch md5 validation files for cached testing data.
• Added a tox modifier for running tests without internet access (tox -e offline)
  • Option to call tox -e offline-prefetch to run the data prefetching command before sockets are disabled in pytest call.
• Added a build using Python3.10 to the GitHub Workflows that uses the new offline modifier in the tox configuration.
• For better accuracy, the tox -e black recipe has been renamed to tox -e lint, as this configuration already included several other linting checks.
• Added a CLI command ($ xclim prefetch_testing_data) for manually calling the populate_testing_data() command with branch (--branch/-b) options.
• Updated the testing data used in the analogs.ipynb notebook to use the testing data now found in Ouranosinc/xclim-testdata's main branch (see: Add analogs notebook data to main branch xclim-testdata#26)

xclim now has a dedicated CLI command for prefetching the Ouranosinc/xclim-testdata:

$ xclim prefetch_testing_data
# or, for a specific testdata branch
$ xclim prefetch_testing_data --branch my_testdata_branch

Running pytest will still perform the fetching of data on first run if the user does not have an xclim-testdata folder in their platform-dependent cache location (/home/$USER/.cache/xclim-testdata on Linux).

Does this PR introduce a breaking change?

Yes:

• tox is now configured to pass positional arguments to the pytest calls directly. This gives developers much more control when selecting/deselecting pytest markers. To replicate previous behaviour (skipping slow tests), most tox invocations should resemble the following:

$ tox -e pyXYZ -- -m "not slow"

• xclim development now requires the pytest-socket plugin (has been explicitly added to the development requirements). This plugin allows developers to mimic being offline when testing new xclim features or bug fixes, if they so choose.

• For better accuracy, the tox -e black recipe has been renamed to tox -e lint, as this configuration already included several other linting checks.

Other information:

https://github.com/miketheman/pytest-socket
https://tox.wiki/en/latest/config.html#substitutions-for-positional-arguments-in-commands

[github-actions]

Note
It appears that this Pull Request modifies the main.yml workflow.

On inspection, the XCLIM_TESTDATA_BRANCH environment variable is set to the most recent tag (v2023.9.12).

No further action is required.

[review-notebook-app]

Check out this pull request on 

See visual diffs & provide feedback on Jupyter Notebooks.

---

Powered by ReviewNB

[huard]

LGTM,
thanks

[coveralls]

coverage: 90.468% (-0.2%) from 90.674% when pulling d80ce92 on offline-testing into 44079ac on master.