Fix parameter name and return type in eleven fix_metadata docstrings

[karpovantonme]

Description

Eleven overrides of Fix.fix_metadata document their argument as cube, but the method takes cubes. Nine of those same docstrings promise an iris.cube.Cube back while the code returns a CubeList.

The rest of the package already gets this right. The base class in esmvalcore/cmor/_fixes/fix.py writes cubes, and so does every other override that documents the argument at all. In gfdl_esm2g.py and noresm1_me.py both spellings live in the same file, so the correct version is sitting a few lines above the wrong one.

Worth saying what this does not do. These modules are not listed in doc/api, so nothing changes on readthedocs. What changes is the source people read: doc/develop/fixing_data.rst asks contributors to write their own dataset fixes, and the natural way to start one is to copy the nearest existing fix.

Docstrings only, no code path touched. In the same files fix_data(self, cube) documents cube: iris.cube.Cube and that one is correct, so I left it alone.

One thing to flag: the guidelines ask for a few files per pull request and this is seven. It is the same one-line change in each and there is no code in it, but I can split it per dataset if that reads better for review.

Link to documentation: https://docs.esmvaltool.org/projects/ESMValCore/en/latest/develop/fixing_data.html

---

Checklist

• 🛠 This pull request has a descriptive title and labels. I cannot set labels from outside the org, documentation is the one that fits
• 🛠 Code is written according to the code quality guidelines. ruff check and ruff format are clean on the changed files
• 🧪 and 🛠 Documentation is available
• 🛠 Unit tests have been added. None: the change is docstring text
• 🛠 Changes are backward compatible
• 🛠 Any changed dependencies have been added or removed correctly. None changed

[CLAassistant]

All committers have signed the CLA.

[valeriupredoi]

this PR is generated by an AI agent/bot (that has links to Telegram, no less), at the moment we do not accept such contributions. Until further decisions, I am putting a hold on this PR.

[karpovantonme]

Understood, your call

Two small corrections for the record. I checked your contributing guide before opening this, and there is no policy on AI assistance in it. The CLA is signed and all checks are green. And I have no Telegram anywhere on my profile or in this PR, so that part may have come from somewhere else.

The change itself is eleven docstrings naming a parameter the method does not take, with the base class right there doing it correctly. If the project would rather wait for the policy to land, I will close it myself and stay out of the repository.

Either way, thanks for the time 🙏🏼