feat: apply-mask (field) filter allows using another field as the mask

[aaron-hopkinson]

Description

Alternative implementation of #276, builds on #274 but converts the filter back to a Filter rather than SingleFieldFilter (but keeps alignment to minimise the diff)

Allow masking from a field in the pipeline

The apply_mask filter currently requires an external file (path) to provide the mask. This makes it awkward when the mask (e.g. lsm) is already available as a field in the same pipeline — you need to save it to a separate file first, then reference it.

This PR adds a mask_param option as an alternative to path. When mask_param is set, by default, the mask field is consumed (not returned), but it is possible to force it to be returned with (return_mask: true). All existing mask_value/threshold/rename/param options work exactly the same way.

Usage

input:
  pipe:
    - mars:
        param:
        - sd
        - lsm
    - apply_mask:
        mask_param: lsm   # Use lsm for masking
        mask_value: 0      # Set to NaN where lsm == 0 (sea points)

As a contributor to the Anemoi framework, please ensure that your changes include unit tests, updates to any affected dependencies and documentation, and have been tested in a parallel setting (i.e., with multiple GPUs). As a reviewer, you are also responsible for verifying these aspects and requesting changes if they are not adequately addressed. For guidelines about those please refer to https://anemoi.readthedocs.io/en/latest/

By opening this pull request, I affirm that all authors agree to the Contributor License Agreement.

[mc4117]

I have tested this PR and checked that the mask works as expected

[mc4117]

Passes all checks. LGTM!

[mpvginde]

Would it be useful to add the option to not remove the field used for masking?
For instance you would want to mask on lsm but also keep it in the final dataset?

[aaron-hopkinson]

@mpvginde Added. Is that okay? 🙂

[mc4117]

I have tested and with the new changes, this transform still works for my use case