Migrate CLI functionality to plugin layer subset 2

[cpelley]

Second/final chunk of plugin changes required by EPP.

Tasks

• improver/cli/extract.py
  • Working acceptance testing.
  • Unittesting coverage.
• improver/cli/freezing_rain.py
  • Working acceptance testing.
  • Unittesting coverage.
• improver/cli/hail_fraction.py
  • Working acceptance testing.
  • Unittesting coverage.
• improver/cli/hail_size.py
  • Working acceptance testing.
  • Unittesting coverage.
• improver/cli/interpolate_using_difference.py
  • Working acceptance testing.
  • Unittesting coverage.
• improver/cli/lightning_from_cape_and_precip.py
  • Working acceptance testing.
  • Unittesting coverage.
• improver/cli/nbhood.py
  • Working acceptance testing.
  • Unittesting coverage.
• improver/cli/phase_change_level.py
  • Working acceptance testing.
  • Unittesting coverage.
• improver/cli/phase_probability.py
  • Working acceptance testing.
  • Unittesting coverage.
• improver/cli/snow_splitter.py
  • Working acceptance testing.
  • Unittesting coverage.
• improver/cli/vertical_updraught.py
  • Working acceptance testing.
  • Unittesting coverage.
• improver/cli/wet_bulb_freezing_level.py
  • Working acceptance testing.
  • Unittesting coverage.
• improver/cli/wet_bulb_temperature.py
  • Working acceptance testing.
  • Unittesting coverage.

Issues

• https://github.com/MetOffice/improver_suite/issues/2079
• Migrate functionality from CLI to plugin layer #1998
• https://github.com/MetOffice/barebones_workflow/issues/32
• https://github.com/MetOffice/improver_suite/issues/2253

[cpelley]

@bayliffe, directly relevant change in reflection of this work in MetOffice/dagrunner#40 (you need not review that, I mention only out of interest).

[cpelley]

This is the type of each positional argument (note the asterisk). That is, each positional argument can be either a cube or cubelist. The documentation should then stipulate that we expect 3 cubes amongst whatever form of input is provided.
I'll flesh out the documentation here 👍

[cpelley]

Changes based on feedback from @bayliffe:

• Moved ignore_error into the plugin of from improver/cli/extract.py.
  • Changes to improver_tests/utilities/cube_extraction/test_ExtractSubCube.py to reflect this change in behaviour to the plugin.
• Recommended doc change to improver.utilities.complex_conversion.deg_to_complex.
• Removed static methods deg_to_complex and complex_to_deg from WindDirection.
• Migrated some testing from improver_tests/wind_calculations/wind_direction/test_WindDirection.py to new file improver_tests/utilities/complex_conversion/test_integration.py (roundtrip testing).
  • Changes to improver_tests/wind_calculations/wind_direction/test_WindDirection.py to ensure successful running of tests that remain.

The only thing I haven't done as I'm not quite following is what the issue is, is in regards to #2003 (comment)
However, I responded to what I think is a misunderstanding. *args is being passed to the function. The idea is just 1 parameter could be passed if its a cubelist, or 2 or 3 etc. I have verified that the documentation does highlight what data is actually needed to be provided.

[cpelley]

You happy to approve @bayliffe?

cheers

[cpelley]

I suspect the CI failures with Sphinx are unrelated to this PR (a result of changes in the conda-forge conda environment, prob. needs something pinning or something unpinning).
Fix for CI in #2011

The base branch was changed.

[cpelley]

I have merged in master into this branch and changed the target (since subset 1 was merged I forgot to change it here).

[cpelley]

@mo-robert-purvis, you mind approving once more (it removed your approval after I changed the target to master).
cheers

[cpelley]

I just merged master into this branch to resolve the CI failure.

action taken

[cpelley]

Thanks both. @bayliffe, I have made the leadtimes optional once more. Thanks for double checking this for me.

[bayliffe]

Thanks Carwyn.